Electron Microscopy of Polymers (Springer Laboratory)

Citation preview

Springer Laboratory

Springer Laboratory Manuals in Polymer Science Pasch, Trathnigg: HPLC of Polymers ISBN: 3-540-61689-6 (hardcover) ISBN: 3-540-65551-4 (softcover) Mori, Barth: Size Exclusion Chromatography ISBN: 3-540-65635-9 Pasch, Schrepp: MALDI-TOF Mass Spectrometry of Synthetic Polymers ISBN: 3-540-44259-6 Kulicke, Clasen: Viscosimetry of Polymers and Polyelectrolytes ISBN: 3-540-40760-X Hatada, Kitayama: NMR Spectroscopy of Polymers ISBN: 3-540-40220-9 Brummer, R.: Rheology Essentials of Cosmetics and Food Emulsions ISBN: 3-540-25553-2 Mächtle, W., Börger, L.: Analytical Ultracentrifugation of Polymers and Nanoparticles ISBN: 3-540-23432-2 Heinze, T., Liebert, T., Koschella, A.: Esterification of Polysaccharides ISBN: 3-540-32103-9 Koetz, J., Kosmella, S.: Polyelectrolytes and Nanoparticles ISBN: 3-540-46381-X Striebeck, N.: X-Ray Scattering of Soft Matter ISBN: 3-540-69855-5 Schärtl, W.: Light Scattering from Polymer Solutions and Nanoparticle Dispersions ISBN: 3-540-71950-2 Khulbe, K.C., Feng, C.Y., Matsuura, T.: Synthetic Polymeric Membranes ISBN: 3-540-73993-7 Michler, G.H.: Electron Microscopy of Polymers ISBN: 3-540-36350-7

Goerg H. Michler

Electron Microscopy of Polymers With contributions by Dr. R. Godehardt ċ Dr. R. Adhikari ċ Dr. G.-M. Kim Dr. S. Henning ċ DP V. Seydewitz ċ DP W. Lebek

123

Prof. Dr. Goerg H. Michler Institut für Physik Institut für Polymerwerkstoffe e.V. Martin-Luther-Universität Halle-Wittenberg 06099 Halle Germany [email protected]

ISBN 978-3-540-36350-7

e-ISBN 978-3-540-36352-1

DOI 10.1007/978-3-540-36352-1 Library of Congress Control Number: 2007942162 © 2008 Springer-Verlag Berlin Heidelberg This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable to prosecution under the German Copyright Law. The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Cover design: WMXDesign, Heidelberg, Germany Typesetting and production: le-tex publishing services oHG, Leipzig, Germany Printed on acid-free paper 987654321 springer.com

Springer Laboratory Manuals in Polymer Science

Editor Prof. Harald Pasch Chair of Polymer Characterization Department of Chemistry and Polymer Science University of Stellenbosch Private Bag X1 7602 Matieland South Africa e-mail: [email protected]

Editorial Board PD Dr. Ingo Alig Deutsches Kunststoff-Institut Abt. Physik Schloßgartenstr. 6 64289 Darmstadt Germany e-mail: [email protected] Prof. Josef Janca Université de La Rochelle Pole Sciences et Technologie Avenue Michel Crépeau 17042 La Rochelle Cedex 01 France e-mail: [email protected] Prof. W.-M. Kulicke Inst. f. Technische u. Makromol. Chemie Universität Hamburg Bundesstr. 45 20146 Hamburg Germany e-mail: [email protected]

Prof. Harald Pasch Deutsches Kunststoff-Institut Abt. Analytik Schloßgartenstr. 6 64289 Darmstadt Germany e-mail: [email protected]

Preface

Electron microscopy and atomic force microscopy have developed into powerful tools in the field of polymer science. By using different techniques and methods, morphological details at length scales from the visible (. mm) up to a few . nm can be detected. Consequently, the microscopic techniques used in polymer research support the tendency, over the last two decades, to shift the level of interest from the μm-scale to the nm-scale region. Systems with at least one structural dimension below  nm are now considered to comprise a new class of materials, the so-called nanostructured polymers or nanocomposites. In addition, the influence of several parameters can be studied by changing the morphology of the material. In particular, the influence of the actual, local morphology on mechanical loading effects can be determined. The micromechanical properties or mechanisms that occur at nano- and microscopic levels form the bridge between structure, morphology and mechanical properties. Therefore, electron microscopy and atomic force microscopy directly contribute to a better understanding of structure–property correlations in polymers. Part I offers an overview of electron microscopy and atomic force microscopy techniques and summarises distinctive applications of polymeric materials. The wide variety of preparation methods used to study polymers with the different microscopic techniques are presented and illustrated with typical micrographs in the chapters of Part II. Each technique is discussed in detail, highlighting its application for solving specific problems arising in the characterisation of materials. The applicability of the microscopic techniques and preparation methods described in Parts I and II to the main classes of polymers is documented in Part III. All relevant groups of solid polymers used domestically, industrially, in research and in medicine are mentioned. The characteristic features and also the variety of structures and morphologies of the different polymer classes are illustrated with typical micrographs. In particular, the application of different microscopic techniques is shown to reveal similar polymeric structures, enabling laboratories that possess only some of the techniques to use them beneficially. As well as descriptions of characteristic morphologies and micromechanical properties the most commonly occurring defects and failures are also illustrated. The volume is directed at polymer scientists from research institutes and industry, and aims to demonstrate the widespread possibilities enabled by the application of microscopic techniques in polymer research and development. Each of these techniques allows one to solve a number of problems, as even for the specialist it is not always evident which technique is best suited to solving a given problem. The mono-

VIII

Preface

graph is also directed at research and applied technicians, since it provides a basic understanding of the principles of the different microscopic techniques and exhausts all of the possibilities of using these techniques to solve specific research problems. All of the preparation methods applied for the study of a variety of polymeric materials using different techniques are described in depth, which will also aid laboratory assistants or students that are new to microscopy, as well as those that wish to improve their skills. Finally, the book will be also helpful for students of polymer physics, chemistry and engineering, as well as those researchers interested in the micro- and nanoscopic world of polymers. This volume draws upon the experiences and studies of the working groups of the editor in research institutes, industry, and academia in the period from  onwards (i.e. over three decades). The authors or coauthors of the various chapters are: Dr. R. Godehardt (Chaps. , , , ) Dr. R. Adhikari (Chaps. , ) Dr. G.-M. Kim (Chaps. , , , ) Dr. S. Henning (Chaps. , , , ) DP V. Seydewitz (Chap. ) DP W. Lebek (Chaps. , , , ) For additional contributions and remarks, I thank Prof. Dr. F.J. Baltá-Calleja, Instituto de Estructura de la Materia, CSIC, Madrid, Dr. W. Erfurth, Max Planck Institut für Mikrostrukturphysik Halle (in Chap. ), Dr. J. Lacayo-Pineda, Continental AG, Hannover (in Chap. ), DI St. Scholtyssek (in Chaps.  and ) and DI M. Buschnakowski (in Chap. ). My former or current coworkers DI (FH) I. Naumann, DI (FH) H. Steinbach, Mrs I. Schülke, Dr. J. Starke, DP J. Laatsch, DI (FH) S. Goerlitz and Mrs C. Becker are gratefully acknowledged for providing many of the examples of microscopic investigations of different polymers and micrographs referred to in this book. I also thank DI W. Schurz for image processing many of the electron micrographs, Mrs B. Erfurt for typing many of the chapters, and DP W. Lebek for his valuable technical help during the completion of the manuscript. Finally, I also wish to gratefully acknowledge the coworkers at Springer-Verlag for their understanding and help during the preparation of the manuscript. Halle/Merseburg, March 

Goerg H. Michler

Abbreviations

General techniques AFM AM BSE CCD CRT EDX(A) EELS EFTEM ELNES EM ESI ESD ESEM FEG FIB FM GIF GSED HRTEM HVTEM ID LFM LVTEM MW MCA MDS MOS PE PEELS PFM ROI SE

Atomic force microscope Amplitude modulation Backscattered electrons Charge-coupled device Cathode ray tube Energy dispersive X-ray analysis Electron energy-loss spectroscopy, electron energy-loss spectrometer Energy-filtered transmission electron microscopy Energy-loss near-edge structure Electron microscope Electron spectroscopic imaging Electron spectroscopic diffraction Environmental scanning electron microscope Field-emission gun Focussed ion beam Frequency modulation Gatan imaging filter Gaseous secondary electron detector High-resolution transmission electron microscope High-voltage transmission electron microscope Interparticle distance Lateral force microscope Low-voltage transmission electron microscope Molecular weight Multichannel analyser Minimum-dose systems Metal-oxide semiconductor Primary electrons Parallel electron energy-loss spectroscopy Pulsed force mode Region of interest Secondary electrons

X

SEM SFM SNOM SPM STEM STM TEM Tg TM TMAFM WD WDX(A)

Abbreviations

Scanning electron microscope Scanning force microscope Scanning near-field optical microscope Scanning probe microscope Scanning transmission electron microscope Scanning tunnelling microscope Transmission electron microscope Glass transition temperature Tapping mode Tapping-mode atomic force microscope Working distance Wavelength dispersive X-ray analysis

Materials/polymers ABS ASA BR AN COC EB EOC EPDM EPR HDPE HIPS iPP LDPE LLDPE MMT MWCNT NBR NR OsO4 PBMA PB PC PCH PE PET PEB PFS

Acrylonitrile-butadiene-styrene Acrylonitrile-styrene-acrylate Butadiene rubber Acrylonitrile Cyclic olefin copolymer Ethylene-butadiene copolymer Ethylene/1-octene copolymer Ethylene propylene diene rubber Ethylene propylene rubber High-density polyethylene High-impact polystyrene Isotactic polypropylene Low-density polyethylene Linear low-density polyethylene Montmorillonite Multiwalled carbon nanotube Acrylonitrile-butadiene rubber Natural rubber Osmium tetroxide Poly(n-butylmethacrylate) Polybutadiene Polycarbonate Polyvinylcyclohexane Polyethylene Polyethylene terephthalate Ethylene-butylene copolymer Poly(ferrocenyl-dimethylsilane)

Abbreviations

PI PMMA PnBA PNC POSS PP PS PTFE PVC PVDF RuO4 PVP SAN SIS SBS

TPE UHMWPE VLDPE

XI

Polyisoprene Polymethylmethacrylate Poly(n-butylacrylate) Polymer nanocomposite Polyhedral oligosilsesquioxane Polypropylene Polystyrene Polytetrafluoroethylene Polyvinylchloride Polyvinylidenefluoride Ruthenium tetroxide Poly(2-vinylpyridene) Polystyrene-acrylonitrile Polystyrene-block-polyisoprene-block-polystyrene Polystyrene-polybutadiene-polystyrene triblock copolymer, polystyrene-block-polybutadieneblock-polystyrene block copolymer Thermoplastic elastomer Ultrahigh molecular weight polyethylene Very low density polyethylene

Table of Contents

INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 3

Part I Techniques of Electron Microscopy 

OVERVIEW OF TECHNIQUES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14



TRANSMISSION ELECTRON MICROSCOPY: FUNDAMENTALS OF METHODS AND INSTRUMENTATION . . . . . . . A Brief History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fundamentals of Electron Optics and Instrumentation . . . . . . . . . . . . .. Some Fundamental Properties of Electrons . . . . . . . . . . . . . . . .. Electron Lenses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Electron-Optical Aberrations and Resolution . . . . . . . . . . . . . .. Vacuum System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Instrument . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Electron Gun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Illumination System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Objective Lens and Specimen Stage . . . . . . . . . . . . . . . . . . . . . . . .. Image-Forming System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Viewing Chamber and Image Acquisition . . . . . . . . . . . . . . . . . .. Alignment and Operation of Transmission Electron Microscopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fundamentals of Image Formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Scattering Mechanism and Contrast Formation . . . . . . . . . . . .. Electron Diffraction and Diffraction Contrast . . . . . . . . . . . . . .. Fundamentals of the Imaging Process . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .



15 15 17 17 19 22 26 27 30 33 35 36 37 39 40 41 44 46 50

TRANSMISSION ELECTRON MICROSCOPY: CONVENTIONAL AND SPECIAL INVESTIGATIONS OF POLYMERS . . . . . . . . . . . . . . . . . 53 . Conventional Investigations Utilising Mass-Thickness Contrast . . . . . 53 . Electron Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

XIV



Table of Contents

.. Selected Area Diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Structure Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . High-Resolution Transmission Electron Microscopy . . . . . . . . . . . . . . . .. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Evaluation and Reduction of Radiation Damage . . . . . . . . . . . .. Application of HRTEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Phase Contrast Transmission Electron Microscopy . . . . . . . . . . . . . . . . .. Phase Contrast at Large Defocus Values . . . . . . . . . . . . . . . . . . .. Phase Contrast by Means of Phase Plates . . . . . . . . . . . . . . . . . . . Electron Holography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Image Plane Off-Axis Holography . . . . . . . . . . . . . . . . . . . . . . . . .. Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Low-Voltage Transmission Electron Microscopy . . . . . . . . . . . . . . . . . . . .. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. A Dedicated Low-Voltage TEM and its Application . . . . . . . . . High-Voltage Transmission Electron Microscopy . . . . . . . . . . . . . . . . . . .. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Advantages and Applications of HVTEM . . . . . . . . . . . . . . . . . . Scanning Transmission Electron Microscopy . . . . . . . . . . . . . . . . . . . . . . .. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Similarities and Differences between STEM and TEM . . . . . .. Application of Bright-Field and Dark-Field Modes . . . . . . . . . Analytical Transmission Electron Microscopy . . . . . . . . . . . . . . . . . . . . . .. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. EELS Investigations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Electron Spectroscopic Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . Electron Tomography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Data Acquisition, Image Alignment and Reconstruction . . . .. Resolution of Reconstructed Data . . . . . . . . . . . . . . . . . . . . . . . . .. Application of Electron Tomography . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

55 55 57 57 57 58 60 60 60 62 62 62 64 64 64 64 65 65 65 66 66 67 68 70 70 72 75 78 78 79 80 81 81

SCANNING ELECTRON MICROSCOPY (SEM) . . . . . . . . . . . . . . . . . . . . . A Brief History of SEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fundamentals of Electron Optics and Signal Generation . . . . . . . . . . . .. Principle of SEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. The Lateral Resolution Power of SEM . . . . . . . . . . . . . . . . . . . . . .. Comparison of Various Cathode Types . . . . . . . . . . . . . . . . . . . .. Depth of Focus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Interaction of Primary Electrons with Sample . . . . . . . . . . . . . . The Instrumentation of SEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. The Column . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Specimen Chamber and Goniometer . . . . . . . . . . . . . . . . . . . . .

87 87 88 88 89 92 92 92 95 95 97

Table of Contents

.. Detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Signal Display and Magnification . . . . . . . . . . . . . . . . . . . . . . . . . . Contrast Formation and Charging Effects . . . . . . . . . . . . . . . . . . . . . . . . . .. Secondary Electron Contrast. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Contrast of Backscattered Electrons (Solid State Detector) . .. Charging Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . X-Ray Microanalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Physical Fundamentals of the Generation of X-Rays . . . . . . . .. X-Ray Microanalysis Techniques . . . . . . . . . . . . . . . . . . . . . . . . . .. Detector for EDX Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Quantitative EDX Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. X-Ray Mapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Wavelength Dispersive X-Ray Microanalysis (WDX) . . . . . . . Environmental Scanning Electron Microscope (ESEM™) . . . . . . . . . . . .. Low-Vacuum SEM and ESEM™ . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Avoiding Charging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. The Wet Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. The Gaseous Secondary Electron Detector (GSED) . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

XV

98 99 100 100 103 103 105 105 107 108 111 112 112 116 116 116 117 119 120



ATOMIC FORCE MICROSCOPY (AFM) . . . . . . . . . . . . . . . . . . . . . . . . . . 121 . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 . Methodical and Instrumental Fundamentals . . . . . . . . . . . . . . . . . . . . . . 124 . Modes of Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 .. Contact Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 .. Force Modulation Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 .. Dynamic Operational Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 .. Tapping Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 . Typical and Special AFM Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142



IN SITU MICROSCOPY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 . Micromechanical In Situ Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 .. Technical Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 .. In Situ Microscopy in (E)SEM . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 .. In Situ Microscopy in (HV)TEM . . . . . . . . . . . . . . . . . . . . . . . . . 150 . In Situ Microscopy in AFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158



IMAGE PROCESSING AND IMAGE ANALYSIS . . . . . . . . . . . . . . . . . . . . 161 . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161 . Image Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162 . Image Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 . Fourier Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

XVI

Table of Contents

. Stereoscopic Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 Part II Preparation Techniques 

PROBLEMS ASSOCIATED WITH THE ELECTRON MICROSCOPY OF POLYMERS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Irradiation Sensitivity of Polymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Low Contrast of Polymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods of Investigating the Morphologies of Polymers . . . . . . . . . . . .. Powders, Particles, Fibres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Bulk Polymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods for Studying Micromechanical Processes . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

175 175 176 177 178 178 178 180 183



PREPARATION OF SURFACES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Surfaces in Their Natural Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Smooth and Etched Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Chemical Etching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Physical Etching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fracture Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Investigation of Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

185 185 185 186 187 191 192 195 196



PREPARATION OF THIN SECTIONS: (CRYO)ULTRAMICROTOMY AND (CRYO)MICROTOMY . . . . . . . . . . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Microtomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Ultramicrotomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Cryomicrotomes and Cryoultramicrotomes . . . . . . . . . . . . . . . .. Knives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Modern Trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Working with a Microtome and an Ultramicrotome . . . . . . . . . . . . . . . . Specimen Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Embedding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Specimen Trimming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Fixation and Staining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ultrathin Sectioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Sectioning Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Wet and Dry Sectioning Techniques . . . . . . . . . . . . . . . . . . . . . . .. Room Temperature Ultramicrotomy . . . . . . . . . . . . . . . . . . . . . .

199 199 200 200 200 200 200 202 203 204 204 205 205 206 206 206 208

Table of Contents

.. Cryoultramicrotomy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Collecting Sections on Grids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Problems, Errors and Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Typical Errors and Possible Solutions . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

XVII

208 209 212 212 212 217



SPECIAL PREPARATION TECHNIQUES . . . . . . . . . . . . . . . . . . . . . . . . . 219 . Preparation of Polymer Films from Solutions . . . . . . . . . . . . . . . . . . . . . 219 .. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219 .. Solution Behaviour of Polymers . . . . . . . . . . . . . . . . . . . . . . . . . . 220 .. Spin-Coating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220 .. Dip-Coating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222 .. Solution Casting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 .. Examples and Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224 . Preparation Using the Focussed Ion Beam Technique . . . . . . . . . . . . . . 227 .. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 .. Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228 .. Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230



PREPARATION FOR (IN SITU) DEFORMATION TESTS . . . . . . . . . . . 231 . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231 . Specimen Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240



CONTRAST ENHANCEMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241 . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241 . Hardening (Fixation) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241 .. Physical Hardening (Fixation) . . . . . . . . . . . . . . . . . . . . . . . . . . . 242 .. Chemical Fixation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243 . Chemical Staining Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243 .. Media Used to Perform Chemical Staining of Polymers . . . . 243 .. Chemical Staining of Compact Specimens Before Sectioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246 .. Chemical Staining of Thin Sections . . . . . . . . . . . . . . . . . . . . . . . 248 . Enhancement of Contrast Through Physical Effects . . . . . . . . . . . . . . . . 248 .. Contrast Enhancement by γ- or Electron Irradiation . . . . . . . 248 .. Straining-Induced Contrast Enhancement . . . . . . . . . . . . . . . . 254 . Problems and Artefacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259

XVIII

Table of Contents

Part III Main Groups of Polymers 

STRUCTURAL HIERARCHY OF POLYMERS . . . . . . . . . . . . . . . . . . . . . . . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Macromolecular Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Constitution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Conformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Macromolecule Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Supramolecular Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basic Relationships Between Macromolecular Parameters and (Micro)mechanical Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

263 263 263 265 265 265 267 269 271



AMORPHOUS POLYMERS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Micromechanical Behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Additional Examples of Amorphous Polymers . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

277 277 277 281 287 293



SEMICRYSTALLINE POLYMERS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Structural Hierarchy in Semicrystalline Polymers . . . . . . . . . . .. Methods of Morphological Analysis . . . . . . . . . . . . . . . . . . . . . . . Micromechanical Behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Brittle Behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Ductile Behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. High-Strength Fibres and Parts . . . . . . . . . . . . . . . . . . . . . . . . . . . Additional Examples of Semicrystalline Polymers . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

295 295 296 296 299 310 310 313 319 321 327



POLYMER BLENDS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Micromechanical Behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Examples. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Blends of Amorphous Polymers . . . . . . . . . . . . . . . . . . . . . . . . . . .. Blends of Amorphous and Semicrystalline Polymers . . . . . . . .. Blends of Semicrystalline Polymers . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

329 329 331 338 341 341 343 343 349

273 276

Table of Contents

XIX



HIGH-IMPACT RUBBER-MODIFIED POLYMERS . . . . . . . . . . . . . . . . . 351 . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351 . Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352 . Micromechanical Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357 . Additional Toughening Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370



BLOCK COPOLYMERS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373 . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373 . Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375 .. Block Copolymer Nanostructures via Self-Assembly . . . . . . . 375 .. Examples of Tailored Block Copolymer Morphology . . . . . . . 379 . Deformation Mechanisms in Block Copolymers . . . . . . . . . . . . . . . . . . . 385 . Special Cases of Self-Assembly and Applications . . . . . . . . . . . . . . . . . . 389 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391

 RUBBERS AND ELASTOMERS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393 . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393 . Morphology of Rubbers and Elastomers . . . . . . . . . . . . . . . . . . . . . . . . . . 395 . Micromechanical Deformation Behaviour . . . . . . . . . . . . . . . . . . . . . . . . 398 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 402 

POLYMER COMPOSITES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405 . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405 . Particle-Reinforced Polymer Composites . . . . . . . . . . . . . . . . . . . . . . . . . 406 . Fibre-Reinforced Polymer Composites . . . . . . . . . . . . . . . . . . . . . . . . . . . 414 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417



POLYMER NANOCOMPOSITES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419 . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419 . Examples of Different Classes of Nanocomposites . . . . . . . . . . . . . . . . . 422 .. Polymer Nanocomposites Based on Zero-Dimensional Filler Particles (POSS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422 .. Polymer Nanocomposites Based on One-Dimensional Filler Particles (MWCNT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 424 .. Polymer Nanocomposites Based on Two-Dimensional Filler Particles (MMT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425 .. Polymer Nanocomposites Based on Three-Dimensional Filler Particles (SiO ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 428



BIOMATERIALS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429 . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429 . Electron Microscopy of Polymeric Biomaterials: Specific Problems and Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 431

XX

Table of Contents

. Examples. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Natural Biomaterials: Bone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. UHMWPE in Orthopaedics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Acrylic Bone Cements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Bioactive Composites for Bone Replacement . . . . . . . . . . . . . . .. Dental Composites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Sutures, Scaffolds and Meshes . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Ureter Stents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Silicone-Based Tracheal Stents and Voice Prostheses . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

432 432 433 434 435 436 437 439 439 443

 SPECIAL PROCESSING FORMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Multilayered Films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Micromechanical Deformation Mechanisms . . . . . . . . . . . . . . . Hot-Compacted Self-Reinforced Polymers . . . . . . . . . . . . . . . . . . . . . . . . .. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanofibres by Electrospinning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Micromechanical Deformation Mechanism . . . . . . . . . . . . . . . . Microformed Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. Several Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

445 445 446 446 446 448 451 451 452 455 455 456 458 460 460 460 462

SUBJECT INDEX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465

Introduction

Polymers form one of the most important groups of materials used in modern industry. The materials in this group encapsulate great diversity in terms of their characteristics, resulting in various polymeric forms such as plastics, rubbers, fibres or dyes. Polymers are applied in almost all sectors of daily life, from households to medicine, agriculture, the automotive industry, up to microelectronics and space research. Polymers are typically composed of very large molecules, known as macromolecules, which usually consist of thousands of repeating units, termed monomers. There is huge variety in terms of the types, numbers, arrangements and combinations of monomers found in polymers, which therefore leads to an extremely wide variety of different polymeric materials. As well as synthetic polymeric materials, there are also natural and biological polymers, including proteins, silk, and cellulose. The wide range of polymeric materials that have good processability, environmental stability, lightweight characteristics and easy machinability make polymers very useful materials. This usefulness is reflected in the worldwide production of all types of polymers, thermoplastics, resins, and rubbers, which has increased enormously since , with annual average growth rates of about % []. This growth continues even now, with growth rates currently about –% per annum, and polymer production reached around  million t in  globally. The production of steel (by volume) was surpassed by polymer production in the year  by about  million m [], in such a way that today polymers are produced in greater amounts than any other group of materials. Steady growth in polymer production is also expected in the near future, based on the abovementioned characteristics and several other developments: better functionality of polymeric parts or a wider range of uses, the same or better performance achieved with lower amounts of raw material (oil) and energy, and better recycling of these materials, thus enhancing sustainable development. It is remarkable that more than % of all polymers are currently based on so-called mass polymers or commodities, e.g. polyethylenes, polypropylenes, polystyrene, polyvinylchloride and rubber. This renaissance of mass polymers is due to improved polymerisation, controlled molecular weight and macromolecular design, better macromolecular regularity, many modifications of the arrangements of known monomers and polymers, and modification with fillers. Examples include the broad fields of polymer blends, high-impact polymers, block copolymers or composites. A shift of interest to smaller and smaller details, from the former μm level to the now increasingly interesting nm level, is currently occurring. This increasing tendency to make structural modifications has also pushed polymer re-

2

Introduction

search to improve morphological control through the use of electron microscopy. There is increasing interest in achieving accurate correlations between synthesis, molecular structure, morphology and properties; see Fig. I.. The most important properties of many applications of polymers include their mechanical properties. Here, the micromechanical processes of deformation and fracture provide the bridge between structure/morphology and mechanical properties. To gain a better knowledge of structure–property correlations, much effort must be expended in studying structures, morphology and micromechanical properties. When used for polymer characterisation, electron microscopic techniques have a tremendous advantage over other methods, as they can provide a direct “view” with a high local resolution of the material of study. Other techniques do not provide direct pictures comparable to those yielded by electron microscopy, but average information about larger volumes instead. With the shift in interest to smaller and smaller structural details, electron microscopy and atomic force microscopy are becoming more and more important. The recent increased availability of high-resolution electron microscopy and atomic force microscopy is now making it possible to view molecular arrangements, and this should lead to further advancements in our understanding of the forms and structures of all types of polymer systems.

Fig. I.1. Correlations between structure, morphology, influencing parameters and mechanical properties of polymers

References

3

An additional trend in microscopy is to reveal not only structural details with improved resolution but also changes in morphology under the action of influencing parameters, such as physical and thermal ageing, outdoor weathering and mechanical loading. In particular, the influence of mechanical loading on changes in the structure and morphology of polymers – in other words, their micro- or nanomechanical properties and mechanisms – can be revealed by electron and atomic force microscopy with an otherwise unattainable accuracy. A better understanding of structure–property correlations enables the defined modification of polymeric structure or morphology and thus the improvement of polymers. While a huge variety of macromolecular and supramolecular structures exist, not all of them are of equal relevance for property improvements. Usually, only a few of the structures dictate the mechanical behaviour of the polymer; these are called “property-determining structures” []. A detailed knowledge of these structures and the underlying micromechanical mechanisms associated with them enable criteria to be defined for the modification and production of polymers with specifically improved or new properties []. This is known as the “microstructural construction of polymers” []. In summary, the technique of electron microscopy plays a decisive role in aiding our understanding of the structure–property correlations encountered in polymer research as well as in the polymer industry.

References 1. Menges G () Werkstoffkunde Kunststoffe. Carl Hanser Verlag, München 2. Editor () Kunststoffe : 3. Michler GH () Kunststoff-Mikromechanik: Morphologie, Deformations- und Bruchmechanismen. Carl Hanser Verlag, München 4. Michler GH, Baltá-Calleja FJ (eds) () Mechanical properties of polymers based on nanostructure and morphology. CRC Press, Boca Raton, FL 5. Michler GH () Polym Adv Technol :

Part I

Techniques of Electron Microscopy

1 Overview of Techniques

This introductory chapter provides an overview of the various techniques of microscopy that are available. Starting with the improvement in the resolution of microscopes during their historical development from optical microscopes up to highresolution transmission electron microscopes and scanning tunnelling microscopes, the field of microscopy is classified based on the principles of imaging. The main techniques used to study the structures of surfaces, the internal structures of polymers and their chemical compositions are listed and discussed in detail in the subsequent chapters in Part I. An overview of additional techniques used to study the morphologies and micromechanical properties of polymers closes the chapter. The structures and morphologies of polymers have been under investigation for more than  years. Early applications of transmission electron microscopy were concerned with the study of the spherulitic crystallisation of natural rubber and lowdensity polyethylene. After  lamellar crystals of polyethylene crystallised from dilute solution were studied by transmission electron microscopy by Keller and Bassett and the folded chain hypothesis was advanced [,]. This was followed by a large number of contributions on the morphology of crystalline polymers. Techniques that allowed the use of transmission electron microscopy to investigate relatively complex structures, such as spherulites, polymer blends and block copolymers, were developed in the s and s. Scanning electron microscopy was introduced in the s, and since then it has been used to investigate fracture surfaces, phase separation in polymer blends and crystallisation of spherulites. The recent availability of high-resolution electron microscopy, coupled with image processing techniques, or atomic force microscopy is now making it possible to view structures down to the molecular level. The primary reason for developing electron microscopes was to improve the resolution of microscopes, and so one of the most important aspects of each microscope is its resolution power, which is the minimum distance between two adjacent object points that can still be imaged separately. It is well known that the resolution of optical microscopes is limited by the wavelength of visible light (it is half of the wavelength, about . μm). After the first development of optical microscopes in the seventeenth century, the improvements made mainly by Abbe, Zeiss and Schott at the end of the nineteenth century yielded microscopes with this resolution and physical evidence of the resolution limit. The development of electron microscopes in the s resulted in  in the creation of the first transmission electron microscope with a better resolution than an optical microscope. Since then there have been enormous im-

8

1 Overview of Techniques

provements in resolution; see Fig. .. Recent advances in techniques used as well as in interpreting and processing the images have allowed resolutions of the order of . nm (=  Å) or better to be achieved for inorganic crystal structures. However, the best resolution achieved in polymers is, in practice, poorer than this because of polymer-specific problems with high electron irradiation sensitivity and low contrast. The next jump in resolution came with the development of scanning tunnelling microscopy or, in general, scanning probe microscopy, which enabled the first ever three-dimensional imaging of solid surfaces with atomic resolution. Scanning probe microscopes do not belong in the field of electron microscopy. However, since they are usually used in close connection with electron microscopy and artefact-free evaluations of structures are easier to achieve when their results are compared with those from electron microscopes, they are discussed in this volume too. Field-ion microscopy, a special surface technique with atomic resolution, is not applied to polymers.

Fig. 1.1. Improvements in the resolution of microscopy

1 Overview of Techniques

9

Electron microscopy (EM) can be divided into the techniques of transmission electron microscopy (TEM) and scanning electron microscopy (SEM). A comparison in terms of resolving power shows that scanning electron microscopes are somewhat intermediate between optical microscopes (OM) and transmission electron microscopes. A significant advantage of using SEM compared to TEM is that the former can image the surfaces of bulk samples with a large depth of focus. This large depth of focus also allows SEM to be used at low magnifications instead of optical microscopes. It is necessary to use all of the microscopic techniques if we wish to study the large variety of morphologies and structures of polymeric materials, i.e. from the sizes and shapes of grains or powders up to crystalline structures; see Fig. .. In general, all of the different types of microscopes can be classified according to whether imaging is

Fig. 1.2. Scheme of possible structures of different sizes present in polymers and resolutions attainable with the different microscopic techniques

10

1 Overview of Techniques

achieved by irradiating the object with a “lamp” or to feeling the surface with a “finger” or “needle” (see Fig. .): . A fixed beam of light or electrons is transmitted through the (thin) specimen (as a transmitted beam) in the transmission mode of the optical microscope and in transmission electron microscopes. . A stationary beam is reflected off the (bulk) specimen surface (as a reflected beam) in the reflection mode of the optical microscopes or in electron mirror microscopes (the latter are not used for polymers). . A focussed beam is scanned across the specimen, passing through the (thin) specimen (scanning transmission EM) or resulting in a reflected beam (as in confocal laser scanning microscopy) or secondary or backscattered electrons in scanning electron microscopes.

Fig. 1.3. Schematic representation of the principles of different types of microscopes (see text)

1 Overview of Techniques

11

. A mechanical tip is scanned across the specimen in order to make use of different physical properties in tunnelling microscopes and atomic force microscopes. When studying the bulk material, the material’s surface or its interior is the target of the microscopic investigations; see Fig. .. The surface can be studied directly with scanning electron microscopy (SEM), atomic force microscopy (AFM) and, indirectly after replication, with transmission electron microscopy (TEM). Ultra- and semithin sections from the interior can be used for TEM and thicker sections for analytical TEM and SEM or for AFM. The traditional electron microscopy technique is stationary-beam TEM, which has been applied to a wide range of materials, including polymers (Chaps.  and ). The main limitation of this approach is that a transparent thin foil that is resistant to damage by the electron beam must be prepared. In addition to this conventional TEM method, special equipment has been developed to achieve high resolution (highresolution TEM, HRTEM), to be able to use high (high-voltage TEM, HVTEM) or low accelerating voltages (low-voltage TEM, LVTEM), for scanning transmission (STEM), for holography and for spectroscopy or emission of X-rays in analytical microscopes (ATEM).

Fig. 1.4. Application of different microscopic techniques to study the surface and interior of a bulk polymeric material

12

1 Overview of Techniques

Scanning electron microscopy (SEM) is currently the most popular of the microscopic techniques (see Chap. ). This is due to the user-friendliness of the apparatus, the ease of specimen preparation, and the general simplicity of image interpretation. The obvious limitation is that only surface features are easily accessible. With SEM, the chemical analysis of different elements is usually possible (energy dispersive or wavelength dispersive analysis of X-rays, EDXA, WDXA). As mentioned above, scanning probe microscopy techniques cannot be classified as types of electron microscopy. However, because of the wide application of atomic force microscopy (AFM) and the fact that it is often used in combination with electron microscopic techniques, it is discussed in Chap. . There are some other techniques of electron microscopy, such as emission EM, mirror-EM, field-electron or field-ion microscopy, which cannot be applied to polymers. In the past, the central reason for using EM was structure and morphology determination, but it is also currently of importance for investigating different processes, i.e. changes in the material caused by interactions with several factors, such as heat, electric or magnetic fields and environmental liquids or gases. Of particular interest is the study of micromechanical processes of deformation and fracture, as discussed in Chap.  (on in situ microscopy). Valuable methods for improving images and quantitatively estimating structures are included in the final chapter of Part I, Chap.  (on image processing). There are a number of reviews and monographs that discuss the different techniques in more detail, e.g. [–], and their application to polymers, e.g. [–]. A good overview of microscopically determined structures present in polymers is provided by A.E. Woodward []. Besides the techniques of EM and AFM discussed in this volume, there are other techniques that are used to study the morphologies and mechanical properties of polymers []. The most important of these techniques are: – Optical microscopy with magnifications of up to about , which is very useful for gaining an overview and as a first step in any morphology analysis – Laser scanning and optical near-field microscopies (scanning beam techniques; see Fig. .), which have improved resolution compared to optical microscopes – Macromolecular orientations can be visualised using optical birefringence; – Acoustic microscopy (ultrasound microscopy) is based on the reflection of ultrasound in the sample, which yields information on density differences, microvoids, cracks, etc., with sizes of less than  μm – Small-angle light scattering (SALS) can be used if the polymers are capable of scattering light due to density or birefringence fluctuations of the order of the wavelength of the light, and is a useful method for studying textures that are larger than about  μm, e.g. spherulites in semicrystalline polymers – Classical methods of light-optical interference are used to detect small details on the order of the wavelength of the light in transparent materials; in particular, the micromechanics of crazes at crack tips in transparent glassy polymers can be investigated

1 Overview of Techniques

13

– Small-angle X-ray scattering (SAXS), the traditional technique used to study periodicities in semicrystalline polymers (e.g. long periods of fibrils or lamellae) and also to detect microcavities (e.g. microvoids between craze fibrils and interfibrillar spacing) – Wide-angle X-ray scattering (WAXS) yields information on the crystallinity (type and size of crystals and lattice defects) in polymers – Rheo-SAXS experiments using X-ray radiation from a synchrotron source allow us to measure in situ changes in structure and crystallinity and to perform realtime measurements at low speeds and frequencies – Small-angle neutron scattering (SANS) characterises the fluctuations in the density, concentration, and magnetic properties of the material and yields information on the conformation, size and mobility of the macromolecular coils, but is far from a routine technique – Infrared (IR) and Raman spectroscopy characterise the type and constitution of the macromolecules present – Rheo-optical methods include a mechanical test performed under static conditions which is carried out simultaneously with optical measurements; among other optical methods, Fourier transform infrared spectroscopy (FTIR) has become one of the most frequently applied tools in rheo-optics, since it enables changes in molecular orientations in polymers and different types of macromolecules in polymer blends to be identified – Dynamic mechanical analysis (DMA) is a very helpful tool for determining relaxation processes, glass transition temperatures and mixing or phase separations of different polymer constituents in blends and copolymers – Differential scanning calorimetry (DSC) measures melting or crystallisation temperatures and degrees of crystallinity – Microindentation hardness gives information on crystallisation behaviour, local mechanical properties, and micromechanical processes – Positron annihilation spectroscopy (PAS) allows us to estimate the local concentration of free volume or the size of nanovoids – Electron spin resonance (ESR) and nuclear magnetic resonance (NMR) measure, for instance, radical formation and macromolecular mobility. The advantage of all of these techniques is that they can be used to investigate larger material volumes and to give integral parameters for the measured structure. On the other hand, their major disadvantage is that they cannot clarify variations in structural or morphological parameters, such as the size distributions of lamellae or particles, orientation differences, local concentrations of additives, deviations in phase separation, and many others. In order to gain an understanding of mechanical properties, in particular strength, elongation at break or toughness, it is not sufficient to measure the “average morphology”, since these properties depend strongly on the variation in structural elements and, for instance, on extreme values of them. Therefore, microscopic techniques with high local resolution are exceptionally important for polymer research and the development of materials with improved mechanical properties. To maximise the information obtained about a particular material, a com-

14

1 Overview of Techniques

bination of electron microscopy (with its high local resolution of structures and variations in details) with some of other integral techniques that provide average values of structures should be utilised if possible.

References 1. Keller A () In: Growth and perfection of crystals (Proc Int Conf Crystal Growth, Cooperstown, NY). Wiley, New York, pp  2. Bassett DC, Frank FC, Keller A () Philos Mag : 3. Glauert AM (–) Practical methods in electron microscopy, vols –. Elsevier Science, Amsterdam 4. Bethge H, Heydenreich J (eds) () Electron microscopy in solid state physics. Elsevier, Amsterdam 5. Williams DB, Carter CB () Transmission electron microscopy: a textbook for materials science. Plenum, New York 6. Amelinckx S, van Dyck D, van Landuyt J, van Tendeloo G () Electron microscopy: principles and fundamentals. Wiley-VCH, Weinheim 7. Goodhew PJ, Humphreys FJ, Beanland R () Electron microscopy and analysis, rd edn. Taylor & Francis, London 8. Zhang X-F, Zhang Z (eds) () Progress in transmission electron microscopy : concepts and techniques. Springer, Berlin 9. Shindo D, Oikawa T () Analytical electron microscopy for materials science. Springer, Tokyo 10. Fultz B, Howe J () Transmission electron microscopy and diffractometry of material, nd edn. Springer, Berlin 11. Li ZhR (ed) () Industrial applications of electron microscopy. Marcel Dekker Inc., New York 12. Sawyer LC, Grubb DT () Polymer microscopy. Chapman and Hall, London 13. Roulin-Moloney AC (ed) () Fractography and failure mechanisms of polymers and composites. Elsevier, London 14. Bassett DC () Electron microscopy and spherulitic organisation in polymers, vol . CRC Press, Boca Raton, FL, p  15. Woodward AE () Atlas of polymer morphology. Hanser, Munich 16. Michler GH () J Macromol Sci Phys B :

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

In this chapter, after a brief history of transmission electron microscopy, the fundamentals of electron optics and instrumentation are described. Following the path of the electron beam, the microscope can be split into the following parts: electron gun, illumination system, objective lens and specimen stage, image-forming system and viewing chamber/image recording. These units are described in detail before the fundamentals of image formation are discussed. The differences between the scattering mechanisms that occur in amorphous and crystalline materials lead to an explanation of image contrast. However, the intensity distribution of the image depends not only on the interaction of the electron beam with the object, but also on the illumination conditions and in particular the action of the objective lens and arranged apertures. This electron-optical imaging process, including microscope aberrations, is described in detail based on the wave-mechanical theory of contrast formation.

2.1 A Brief History Recently, key events in the history of electron microscopy have been documented extensively [], and a few reviews [–] have been published that are excellent sources of individual reminiscences and further information concerning the origins and the historical development of electron optics and electron microscopy. The year that saw the birth of the transmission electron microscope (TEM) appears to be a little vague. In  Knoll and Ruska published their results obtained with magnetic lenses and by applying two-step imaging in Zeitschrift für Physik in a rather detailed form, and they designated one chapter of their paper “Das Elektronenmikroskop” (see, e.g., []), and so  is usually said to be the year that the electron microscope was invented. A few years before this, the investigations of Busch had shown that rotationally symmetrical magnetic and electric fields possess some characteristics of lenses, and that the simple lens equation known from optics holds when imaging with such lenses. In  Brüche and Scherzer published the first monograph on geometrical electron optics. The first images from an electron microscope were demonstrated in , and in  the first reliable images in which the resolution of the optical microscope had been surpassed were made available. Soon after, in , Scherzer demonstrated that the spherical and chromatic aberration coefficients of electron lenses are intrinsically nonvanishing, and hence cannot be eliminated by skilful design. Within the next three years, TEMs were developed by commercial companies

16

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

and electron microscopy was established in countries other than Germany. Furthermore, the first scanning transmission electron microscope (STEM) was constructed by von Ardenne in . This pioneering period that laid the foundations for electron microscopy ended with the beginning of World War II. In the second period, which lasted to the end of the s, the number of electron microscopists increased enormously and TEMs became widely available from several sources. Both the preparation of the specimens to be investigated and the electron microscopes were improved significantly. The principle of the stigmator was developed by Bertein (), Hiller and Ramberg () and Rang (). The wobbler, a focussing aid, was invented by Le Poole in . Almost all the methods of correcting the spherical aberration of electron lenses were listed by Scherzer in , and the first studies on image formation and resolving power in wave-optical terms were carried out by Scherzer in  and Glaser in . The use of an electron biprism to obtain interference fringes was demonstrated by Düker and Möllenstedt in . The third period, which covers the time from the late s right up to the present, brought further continuous improvement in all of the operational systems of the TEM as a result of technological progress in general. In particular, progress in electronics combined with the replacement of vacuum tube technology by solid-state circuitry and later on by integrated circuit technology was a driving force for significant improvements in the critical parts of the TEM, such as the power supplies for the lens current and the high voltage. Furthermore, important improvements were made in the objective lens, such as the use of lenses exploiting superconductivity (first proposal by Laberrigue and Levinson in ), the introduction of the condenser objective lens (Riecke and Ruska in ), and the introduction of the high-quality multipurpose objective lens (Mast et al. in ). This initially led to the development of high-resolution TEMs (HRTEMs) with a voltage of about  kV in the s. Then, in the s, instruments capable of a resolution of less than . nm became available for the intermediate voltage range of – kV. On the other hand, interest in high-voltage TEMs (HVTEMs) increased. In  Dupouy published the first pictures he had obtained with the . MV TEM, constructed in Toulouse. Ten years later, the same author presented the first results obtained with the Toulouse  MV TEM. Further high-voltage projects in Cambridge and Japan were completed in the middle of the s. In the s, it became possible to construct HVTEMs with sufficient accelerating voltage stability to allow investigations to be performed at a resolution of . nm, as shown by a  kV TEM from JEOL working at the Max Planck Institute in Stuttgart []. Furthermore, the range of applications of TEMs could be significantly extended by adding supplementary equipment for microanalytical investigations. The first description of an analytical microscope combining a TEM and a microprobe analyzer was given by Duncumb in . Although, this “EMMA” was not a commercial success, it was the first step towards the next generation of analytical TEMs based on energy dispersive X-ray (EDX) analysis. One year later in , a post-column electron energy-loss spectrometer (EELS) for use with a TEM was presented by Wittry, and the development of parallel EELS (PEELS) attachments was started in  following results reported by Jones et al. Besides these external attachments for analytical

2.2 Fundamentals of Electron Optics and Instrumentation

17

investigations, the implementation of an in-column imaging electron energy filter was an important step forward in analytical electron microscopy. In this context the Ω-filter introduced by Rose and Plies in  was a milestone in the development of the energy-filter TEM (EFTEM). Furthermore, the development of STEM was an especially important milestone in the evolution of analytical electron microscopy. Crewe et al. presented a high-resolution STEM based on the first major use of a fieldemission gun in . The first basic study of the mechanism of image formation in the STEM was carried out by Thomson and Zeitler in , and Crewe presented images of single atoms obtained by Z-contrast in the same year. High-quality STEMs have been available commercially since . Revolutionary developments in computer techniques have also considerably improved the design, adjustment and working routines of TEMs since the start of the s, and computer-assisted and computer-controlled microscopy has become the state of the art. Furthermore, due to the development of high-resolution digital recording media, such as image plates and slow-scan CCD cameras, digital image recording and processing have become increasingly popular. One offshoot of this development is improved data acquisition for a tilt series, which provides the basis for the three-dimensional reconstruction of an object by electron tomography. Originally pioneered in the life sciences, this technique has recently become a powerful imaging and analytical tool in materials science [–]. Furthermore, since the end of the s the doors have opened to the new aberration-corrected world of both conventional and scanning transmission electron microscopy [–]. The breakthrough came with the use of non-round lenses. The principle for this had already been outlined by Scherzer in , as mentioned above. These multipole lenses are capable of generating a negative value of the spherical aberration coefficient Cs which can then cancel the positive Cs of the round lens. The resolution of a spherical aberration-corrected TEM or STEM is limited by the energy spread of the incident electron beam and by the uncorrected chromatic aberration. Schemes for the correction of the latter have been proposed by Rose []. On the other hand, the energy spread of the primary beam can be significantly reduced by using a cold field-emission gun, and additional reduction is possible with the aid of a monochromator. Moreover, the application of a monochromator allows highresolution EELS spectroscopy. Very recently, a new corrected (S)TEM platform has been developed that is capable of the highest TEM and STEM performance; it permits a lateral resolution of far better than . nm and an energy resolution of down to . eV to be attained [, ].

2.2 Fundamentals of Electron Optics and Instrumentation 2.2.1 Some Fundamental Properties of Electrons Although Stoney had already introduced the term electron to designate an elementary charge more than ten years before, it was not until the experiments of Thomson in  that the term electron was used with its present-day meaning. Experiments

18

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

like those performed by Thomson to measure the ratio of charge to mass (em− ) for the electron, as well as the series of famous “oil drop” experiments of Millikan in  to determine the electronic charge e, showed incontrovertibly that electrons act like particles. On the other hand, de Broglie stated the principle of the wave-like nature of matter in , as reflected in the famous relationship λ=

h , p

(.)

where λ is the associated wavelength, h is Planck’s constant and p is the magnitude of the particle momentum p. This relation between the particle nature and wave-like nature of matter can also be expressed in the following form p = hk .

(.)

Here, k is the wave vector, the magnitude of which can be written as k = k =

 . λ

(.)

In the case of the electron, the physical existence of the matter wave was demonstrated by Davisson and Germer in  in the first electron diffraction experiment. An electron passing across a large potential difference V is accelerated to a velocity v, which may well approach the velocity c of the light in vacuum. Therefore, relativistic effects must be taken into account. With m denoting the rest mass of the electron, the relativistic change of the electron mass m in relation to the velocity v is given by the well-known equation m=

m  .  − vc 

(.)

The change in energy of the electron caused by its transit across the potential difference V can be expressed by mc  = m  c  + eV .

(.)

The momentum p is estimated by combining Eqs. . and .: 

eV   p = mv = eVm  +    . c

(.)

Thus, using Eq. ., the wavelength of the electrons depends on the potential difference, or in other words on the accelerating voltage in the following way: eV  h λ = = h eVm  +    p c

− 

.

(.)

Using the fundamental constants given in Table ., the properties of electrons as a function of the accelerating voltage are represented in Table ..

2.2 Fundamentals of Electron Optics and Instrumentation

19

Table 2.1. Fundamental constants Charge of the electron (−e) Rest mass of the electron (m 0 ) Velocity of light in vacuum (c) Planck’s constant (h)

−1.602  10−19 C 9.109  10−31 kg 2.998  108 m s−1 6.626  10−34 N m s

Table 2.2. Electron properties as a function of accelerating voltage Accelerating voltage Velocity of electrons Ratio of electron velocity to light velocity vc − V [kV] v [ms− ] 10 100 200 500 1000 3000

5.83  107 1.64  108 2.08  108 2.59  108 2.83  108 2.97  108

0.194 0.548 0.695 0.863 0.941 0.989

Wavelength of electrons λ[pm] 12.20 3.70 2.51 1.42 0.87 0.36

2.2.2 Electron Lenses In electron optics (see, e.g. [–]) we use the same principles as used in traditional light optics to describe the formation of images by lenses, and we also use corresponding terms to characterise the action and aberrations of a lens. The ray diagrams in Figs. . and . show two fundamental features of image formation by an ideal lens, where the lens is assumed to be a so-called “thin” lens (which means it is thin enough that its action on the paths of electron rays through the lens can be illustrated by refractions of the electron rays in the principal plane of the lens). In Fig. . a so-called “self-luminous object” is assumed, and the radiation is emanating from an off-axis point in the object plane. Based on the action of apertures in an electron microscope, a limiting diaphragm restricts the angular spread of the electrons entering the lens. All electron rays passing through the lens are refracted in its centric plane in such a way that they form a point image at their crossover. This is the first fundamental action of an ideal lens. Only two of all of the electron rays that meet at the image point are needed to find the location of this point when drawing a ray diagram. Fortunately, we use two special ray paths for this purpose, as shown by the two bold rays in the figure. The first one is a ray passing through the lens at its centre, the direction of which is not changed by the action of the lens. The second is a ray with a path that is parallel to the electron-optical axis before entering the lens. This one is refracted by the action of the lens in such a way that it crosses the electron-optical axis at the focus in the back focal plane of the lens. In Fig. ., the image formation of a finite object (represented by an arrow) is illustrated. Starting at three different points in the object plane (at the point, at the middle and at the end of the arrow), in each case a set of three electron rays are drawn in the ray diagram. The set of rays at each starting point consists of a ray parallel to the electron-optical axis and two rays with the same inclination to the latter. Following Fig. . we find, for each of the selected starting points in the object plane, a corre-

20

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

Fig. 2.1. Formation of the image of a point source. A diaphragm restricts the angular spread of electrons entering the lens and contributing to image formation. The crossover of two special ray paths is used to find the image of the point object

sponding image point by finding the crossover point for the set of three rays in the image plane. The image plane is conjugate to the object plane, which means that the planes are electron-optically equivalent, and thus rays leaving a point in one plane are brought to a point in the conjugate plane and vice versa. Using the distances u and v (as labelled in Fig. .), the magnification M of the imaged object can be expressed by M=

v , u

(.)

and, introducing the focal length f of the lens, Newton’s well-known lens equation    + = u v f

(.)

is valid, as described in standard textbooks on light optics. Figure . also illustrates the second important feature of image formation by a lens: parallel rays starting at different object points are focussed in the back focal plane of the lens, and the distance of this focus from the electron-optical axis increases with increasing oblique incidence of the parallel beam. The special case of rays entering the lens parallel to

2.2 Fundamentals of Electron Optics and Instrumentation

21

Fig. 2.2. Illustration of the two fundamental features of image formation using a ray diagram for a finite object (represented as an arrow). The three important distances in the diagram are labelled u, v and f . All rays emerging from a point in the object plane are gathered by the lens and converge to the conjugated point in the image plane. On the other hand, all parallel rays starting at different points in the object plane are focussed in the back-focal plane of the lens

the electron-optical axis results in a focus that is positioned on this axis; this is known as the focus of the lens. The Lorentz force which a moving electron experiences in electric and magnetic fields and the resulting deflection of the electron provide the physical basis for electron lenses. With an electric field strength E and a magnetic flux B, the Lorentz force F is F = −e(E + v  B)

(.)

where −e and v are the charge and velocity of the electron, respectively. It is worth noting that the magnetic field action expressed by the vector cross-product of v and B results in a force vector that is normal to v and B. Inserting Eq. . into Newton’s law of motion m¨r = F

(.)

yields the law of particle optics. Although one can in principle use either electrostatic or magnetic lenses to focus a beam of electrons, magnetic rather than electrostatic lenses are preferred because they are more convenient to use and have lower aberrations, so only magnetic

22

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

lenses will be considered here. A homogeneous magnetic field already acts as a weak electron lens for rays with a small inclination with respect to the field direction. For practical purposes, however, magnetic lenses with short focal lengths are obtained by concentrating the magnetic field by means of pole pieces. As illustrated in Fig. ., a conventional magnetic lens consists of a coil of copper wire wound symmetrically around the electron-optical axis. The coil is enclosed by a soft iron shield apart from a narrow gap between a pair of pole pieces across which the focussing field appears. The lens is energised by passing current through the windings, and the magnitude of this current determines the strength of the lens expressed by the focal length. The focussing action of the lens arises as follows. Electrons travelling initially parallel to the electron-optical axis experience upon entering in the field a tangential force due to the interaction of the axial velocity with the radial component of the magnetic field. Therefore, the subsequent direction of travel of the electrons is inclined to the electron-optical axis, and the tangential velocity component interacts with the axial component of the field to produce a radial force toward the axis. The resulting electron paths are helical and, as the field action becomes stronger with increasing distance from the electron-optical axis, a crossover of the electron paths in the lens focus results. An image rotation (marked by the angle ϕ in Fig. .) caused by the helical paths of the electrons within the focussing magnetic field is a typical feature of the action of a magnetic lens. 2.2.3 Electron-Optical Aberrations and Resolution In addition to the simple image formation discussed before, a more realistic one must take into account electron-optical aberrations. Unfortunately it is not possible to can-

Fig. 2.3. Concentration of a rotationally symmetric magnetic field in the gap between a pair of lens pole pieces and its principle action on an electron beam

2.2 Fundamentals of Electron Optics and Instrumentation

23

cel out or correct aberrations in electron optics simply by combining positive and negative elements of different refractive indices, as is done in light optics. Instead, it is necessary to choose operating conditions such that the influence of aberrations is minimised. The aberrations arise, on the one hand, from aberrations of the lens itself and, on the other hand, from the so-called diffraction error, which is a consequence of the presence of diaphragms. Aberrations limit the resolution or resolving power of a microscope, which is the ability to make out points which are close together in the object seen in the image. In the following, only a simplified method is used to describe the resolving power of a TEM. It is related to the image of two adjacent self-luminous object points, the radiations from which are completely independent of each other (incoherent illumination). A single point source will not be imaged as a point, even when no lens aberrations are present. The finite size of the lens results in the diffraction of the rays at the outermost collection angle of the lens, usually defined by a limiting aperture. This diffraction results in a point being imaged as a disc, usually called an Airy disc since the Airy function describes its intensity profile. The Rayleigh criterion is used to define the closest distance between two object points for which the points are still distinguishable in the image. The Rayleigh condition is fulfilled when the maximum from one source lies in the first minimum of the other source. Under these circumstances the distance between the two incoherent point sources is defined as the theoretical resolution r th of the lens and is given by the radius r d of the Airy disc r th = .

λ α

(.)

where λ is the wavelength of the imaging electrons and α is the semi-angle of collection of the lens (compare Fig. .). As aberrations can always be expected in electron microscopes, it is necessary to modify the resolution derived solely for diffraction effects shown in Eq. .. Reimer classified lens defects by ten kinds of lens aberrations []. In practice, however, only spherical aberration, chromatic aberration and astigmatism limit the microscope performance substantially. Because of the technical difficulties involved in correcting electron lenses, one must use electron beams with very small aperture angles in order to avoid image disturbances. Since the object region imaged in the TEM is extremely small, the imaging electron rays are likewise only small distances from the electron-optical axis, so that out of all of the geometrical aberrations only the spherical aberration, which is independent of the distance of the ray from the electron-optical axis, has to be taken into account. Compared to other geometrical aberrations, which depend on the distance from the electronoptical axis in a linear (coma), quadratic (off-axis astigmatism) or cubic (distortion) manner, only the spherical aberration causes a blurring of the image point on the electron-optical axis, thus limiting the achievable resolution. Spherical Aberration Spherical aberration is the inability of a lens to focus all incident rays from a point source to a point. This defect is caused by the lens field acting inhomogeneously on the off-axis rays. The effects of spherical aberration are shown in Fig. ..

24

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation Fig. 2.4. Formation of a disc of confusion due to spherical aberration

For reference we define the true or so-called Gaussian image plane as the image plane for paraxial imaging conditions. In paraxial imaging conditions, the rays are near the electron-optical axis and make only small angles with respect to the electron-optical axis. On the other hand, Fig. . shows that the further off-axis the electron is, the more strongly it is bent back toward the electron-optical axis and its focal length decreases. Therefore, an image disc is obtained of an object point in the Gaussian image plane. The radius r s of the aberration disc in the Gaussian image plane depends on the aperture angle α  , and is expressed by r s = Cs α  .

(.)

The quantity Cs is the coefficient of spherical aberration, which is a characteristic constant with the dimensions of a length that determines the quality of an electron lens. A value of about  mm is typical of Cs for an objective lens in a TEM, while in a HRTEM the value is lowered to about  mm. For an electron lens arranged behind the objective lens, the aperture of the incident ray is smaller than the aperture of rays incident into the objective by a factor given by the magnification of the objective lens. For this reason only the spherical aberration of the objective lens needs to be taken into account when evaluating the resolution limit of the microscope. Chromatic Aberration The other aberration that affects the performance of an electron microscope is chromatic aberration. By analogy with the corresponding effect in light optics whereby the focal length of a lens varies with the wavelength of the light, the focal length of

2.2 Fundamentals of Electron Optics and Instrumentation

25

an electron lens varies with the energy of electrons. The lens bends electrons of lower energy more strongly, and thus electrons from a point in the object once again form a disc image in the Gaussian image plane. The radius r c of this disc is given by rc = Cc

ΔE α , E

(.)

where C c is the chromatic aberration coefficient of the lens (like Cs it is a length), ΔE is the deviation of the electron energy from its mean value E, and α  is again the aperture angle of the lens. The chromatic aberration coefficient C c of a magnetic objective lens is usually slightly smaller numerically than the focal length. Chromatic aberration is a lens defect that degrades the image whenever electrons in the beam cease to be monoenergetic. This may be the result of electrons starting from the gun with a spread of energies, or of the accelerating voltage fluctuating with time, or of the electron beam losing energy through collisions when passing through the specimen. In modern instruments the stability of the accelerating voltage and also that of the lens current, which has a similar influence on chromatic aberration, is so good that we don’t have to worry about the chromatic aberration caused by the illumination system, as it is insignificant when compared with the energy losses associated with the electrons that are transmitted through a sample. Inelastic scattering of the high-energy electrons by plasmon excitations is a common way for electrons to lose – eV, and for thick samples the energy-loss spectrum is additionally broadened by multiple energy losses. A rule of thumb provided by Sawyer and Grubb is that, for biological and polymeric specimens, the resolution limit is about one-tenth of the specimen thickness []. Therefore, thin specimens have to be used to minimise the blurring of TEM images caused by chromatic aberration. Astigmatism Astigmatism is another lens defect that can degrade the resolution of an electron lens. This defect occurs when a lens does not have perfect cylindrical symmetry. It arises from lack of perfection in the machining of the lens pole pieces, in particular a lack of circularity in the bores and the flatness of the pole faces, and also from asymmetry in the magnetic material of the lens itself. The characteristic feature of astigmatism is that beams leaving the object in two perpendicular planes (containing the electron-optical axis) intersect in different image planes. The difference between the focal lengths of these planes that lie perpendicular to each other is used to measure the astigmatism. Fortunately, astigmatism is one of the few defects in electron lenses that can be corrected. Two lenses of the TEM require routine corrections for astigmatism using a “stigmator”. On the one hand, the first condenser lens must be stigmated to produce a circular incident beam on the specimen. On the other hand, an objective stigmator is necessary to cancel the objective astigmatism in order to get better resolution. A stigmator introduces a balancing cylindrical lens field perpendicular to the astigmatic effect of the lens and hence compensates for this effect. The stigmator can be either electromagnetic or electrostatic in nature, the essential requirements being that the magnitude and direction of the compensating field should be independently variable.

26

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

Given a thin specimen and ideal conditions during the operation of the microscope, such as stable object position, adequate alignment of the electron-optical system, accurate focussing, etc., the ultimate resolution limit attainable in electron microscopy depends, on the one hand, on the diffraction aberration and, on the other hand, on the spherical aberration of the objective lens. Since according to Eq. . the diffraction aberration decreases with increasing aperture angle, and according to Eq. . the spherical aberration increases with the third power of the aperture angle, we would like to find the optimum aperture angle at which the combined aberration caused by diffraction and spherical aberration has a minimum. If we take the combination of the diffraction and spherical aberration discs in quadrature

 .λ   r(α  ) = r d + r s =    + (Cs α  ) , (.) α the compromise value exists when dr − .λ  α − + Cs α  = . = dα  r

(.)

From this equation the optimum (compromise) value of the aperture is obtained as 

α ,opt



λ  λ  = .   = A    . Cs Cs

(.)

If this expression for α ,opt is substituted into Eq. ., we can calculate a minimum value for the theoretical resolution limit rth,min : 



rth,min = . Cs λ    = A  Cs λ    .

(.)

Because of the only semi-quantitative significance of Eqs. . and ., the numerical values should be replaced by the constants A  and A  , as then the equations also correspond to the results of wave-mechanical calculations, which lead to somewhat different numerical values (compare Sect. .., Eq. .). The value for rth,min is typically about .–. nm, but for high-resolution instruments it decreases to about . nm. 2.2.4 Vacuum System The vacuum system of an electron microscope is necessary for two reasons. On the one hand, it is essential to remove most of the air molecules from the column of the microscope in order to minimise the scattering of the electron beam by gas molecules so that the electrons can travel from the gun to the specimen and from the specimen to the viewing screen or camera. Considering how the mean free path of the electrons depends on the reduced pressure, this requirement is easily met by using a relatively

2.3 The Instrument

27

modest vacuum system that provides pressures of about − Pa or better in the microscope column. On the other hand, a vacuum of sufficiently low pressure is needed to prevent the specimen, the apertures and the parts of the electron gun becoming contaminated by contaminants such as hydrocarbons and water vapour created by the interaction of the electron beam with molecules of the residual gas. In general, a better vacuum results in lower contamination. Since a very good vacuum is not needed in all parts of the microscope, the present trend is towards a localised higher and cleaner vacuum of about − Pa, as provided by sputter ion pumps, in the specimen chamber to reduce the contamination of the sample and in the electron source chamber for operating LaB sources or field-emission sources. In addition to the action of the vacuum pumps, the vacuum around the specimen is usually improved by applying an internally supplied cryo pump in the form of metal parts close to the specimen that are cooled by liquid nitrogen. By applying differential pumping the ultrahigh vacuum parts of the microscope can be separated in a dynamic way from the high vacuum parts by apertures of appropriate diameters. Generally, the operating vacuum in electron microscopes is provided by a cascade system of different pumps, as each different types of pump only works over a limited vacuum range. The principles and features of the vacuum pumps employed in electron microscopes are summarised in Table .. A rough vacuum is achieved with a rotary mechanical pump, and an oil diffusion pump backed by the rotary pump is usually used to achieve a working pressure of about − Pa. Other types of pumps sometimes used include turbo molecular pumps, sputter ion pumps and cryo pumps, which have the advantage of giving a cleaner vacuum, in particular less contamination from hydrocarbons. Each operational step of the complex vacuum system of an electron microscope is realised by performing special pumping sequences of the different pumps combined in the system. In practice, however, the user rarely needs to be aware of these pumping arrangements since the pumping sequence is automatic and safety devices ensure that the microscope cannot operate until an appropriate vacuum is reached.

2.3 The Instrument Although the basic principles of operation of a TEM have not changed since the beginning of electron microscopy,  years of electron microscopic development has resulted in a lot of improvements and refinements, such that modern instruments really are wonders of technical innovation. Figure . shows the appearance of a modern conventional TEM. It is additionally employed with special attachments for analytical investigations, which are discussed in Sect. .. The arrangement of basic components in a TEM is illustrated in Fig. .. Electrons emitted from a thermionic or a field-emission source are accelerated in the gun by a high voltage produced via a high-voltage generator. The electron beam is formed with the aid of condenser lenses and sometimes a condenser mini-lens (if the objective lens is a condenser/objective lens), a condenser lens aperture, a condenser lens stigmator and beam tilt and translate coils for alignment, and then it enters the objective lens and strikes the specimen in appropriate ways. The specimen is usually

28

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

Table 2.3. Principles and features of vacuum pumps employed in electron microscopes (from [34]; reproduced with permission from Springer) Vacuum pump

Principles

Rotary pump (RP)

Atmosphere to As the pump works at atmospheric pressure, it is employed for rough 10−2 Pa pumping of the TEM. It is also used for back-line evacuation of oil diffusion pumps (DP) and turbo molecular pumps (TMP). Whenever the pump stops, the pump chamber should be set to atmospheric pressure to prevent the oil from flowing back. An oil vapour jet is 10−1 to 10−8 Pa As the pump works at a lower vacuum level than the RP and has a high evacueffused from a nozzle ation speed, it is used to evacuate by heating oil. Gas large-capacity camera chambers in molecules are swept which much gas is generated. Inaway with the oil jet. creased the back-line pressure (decreasing the vacuum) streams oil vapour back into the fore-line. Therefore, the back-line should be evacuated continuously with an RP. Ions generated by 10−2 to 10−9 Pa As the pump is oil-free, it is called a dry pumping system and is used magnetron discharges to evacuate electron gun chambers are sputtered onto the and columns. Because it adsorbs residsurface of a titanium ual gas, it is not suited to areas with wall. The active molelots of residual gas. It is better used cules generated trap to maintain high vacuum in the system. gas molecules, which It is impossible to adsorb inert gas are adsorbed on the molecules, such as helium and argon, thin wall. using an SIP. The pumping power is recovered by maintenance, during which the pump is baked and evacuated by a DP. 10−2 to 10−8 Pa As the pump works in the range from Gas molecules are low to high vacuums and is oil-free, evacuated by rotating it is used to evacuate columns. To a metal rotor fin avoid vibrations, a magnetic buoyantat high speed. type rotor is employed. The back-line is evacuated by an RP. The pump adsorbs all gas molecules, Gas molecules are ad- 10−2 to including inert gases. It is possible sorbed on the surface 10−13 Pa to attain an ideal vacuum. An antiof a metal fin cooled contamination fin installed in the with a coolant such as specimen chamber is considered to be liquid nitrogen. a kind of cryo pump.

Oil diffusion pump (DP)

Sputter ion pump (SIP)

Turbo molecular pump (TMP)

Cryo pump (CP)

Pump sucks, compresses, and evacuates gas by rotating a rotor in a chamber kept sealed and lubricated with oil.

Working vacuum range

Features and notes

2.3 The Instrument

29

Fig. 2.5. The LEO 912 OMEGA, which has an imaging energy filter, is an example of a modern TEM used for polymer investigations

placed via the specimen holder and the goniometer stage within the objective lens between the upper and the lower pole pieces. After passing through the specimen, the electrons form an image through the action of the objective lens and an objective aperture in the back focal plane of the lens. The image is corrected by an objective stigmator and enlarged by an image-forming system consisting of a series of intermediate and projector lenses and alignment units, and thus finally a highly magnified image becomes visible on the viewing screen or can be recorded by a camera or an electron-sensitive film. Following the path of the electron beam, the microscope can be split into the following parts: electron gun, illumination system, objective lens and specimen stage, image-forming system and viewing chamber/image recording. These units are described in the following sections.

30

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

Fig. 2.6. Basic components of a TEM

2.3.1 Electron Gun The electron gun is located at the top of the microscope column. It generates electrons at the negatively biased cathode and accelerates them between the cathode and anode in such a way that the paths of the accelerated electrons form a crossover which acts as a virtual electron source for the first condenser lens of the illumination system discussed in the next section. For many years a sharply bent tungsten wire, as shown in Fig. .a, was widely used as the electron source in conventional TEMs. When heated electrically, this tungsten hairpin filament gets hottest at its sharp tip, where the highest resistance to current flow is found. If the temperature is high enough, some electrons receive sufficient thermal energy to surmount the work function of the tungsten/vacuum interface and leave the tungsten cathode. Increasing the temperature also increases the thermionic emission, but unfortunately also leads to evaporation of the filament material and a decrease in filament lifetime. The conventional thermionic emission from a tungsten hairpin cathode is limited in terms of temporal coherence by an energy spread of the emitted electrons of the order of a few eV

2.3 The Instrument

31

Fig. 2.7. Various electron sources [tungsten filament (a), LaB6 emitter (b) and tungsten field-emitting tip (c)] and schematic diagrams of a thermionic gun (d) and a field-emission gun (e)

(electronvolts) and in terms of spatial coherence by the gun brightness. Brightness, defined as current density per unit solid angle of the source, is an important property of the electron source. A smaller source size gives higher brightness and better spatial coherency, but often less stability. Therefore, cathodes of LaB (or other borides) and field-emission cathodes provide alternatives that yield reduced energy spread and increased gun brightness. The lower work function of LaB has made it the preferred material for thermionic electron sources. This lower work function more than overcomes its lower operating temperature compared to tungsten. LaB cathodes consist of small pointed crystals, as shown in Fig. .b. They require indirect heating because their electrical resistance is too high for direct current heating. LaB crystals are more susceptible to thermal shock than tungsten, so it is important to take care when heating and cooling an LaB source, and as LaB is a highly reactive material, the gun vacuum has to be − times better than that for tungsten cathodes. As shown in Fig. .d, a thermionic electron gun is a triode system consisting of the cathode connected to a highly negative potential, the (grounded) anode aperture, and the Wehnelt electrode in-between, which has a potential that is hundreds of volts more negative than the cathode potential. Therefore, electrons emitted from the point of the electrically heated cathode are not only accelerated towards the anode but are

32

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

also strongly influenced by the action of the Wehnelt electrode. The negatively biased Wehnelt electrode serves to restrict the electron emission, and also serves to focus the electrons to a crossover as they are accelerated to the anode. Usually the Wehnelt electrode is “self-biased” electrically, as the electron current in the microscope flows through the bias/emission potentiometer. This has the added effect of stabilising the emission current against fluctuations in filament temperature, since any tendency for the emission to increase results in a more negative bias voltage which acts against the increase. Several values of bias resistors are normally provided and can be switched in and out according to how much current is required from the gun. In a field-emission gun (FEG), the thermionic emission is replaced by electron extraction through quantum tunnelling. Under ultrahigh vacuum conditions, a material subjected to a sufficiently strong electric field will emit electrons in the region of maximum field strength. A cold FEG employs tungsten with a () plane surface as an emitter, which works at room temperature without heating. As shown in Fig. .c, such a tungsten cathode is fabricated with a sharply pointed shape and a  nm radius of curvature to localise the electric field. As well as a cathode with a sharpened tip, an arrangement of two anodes is used in a FEG, as illustrated in Fig. .e. The first anode is positively biased by several kV with respect to the cathode. This extraction voltage V generates the field strength at the cathode that pull electrons out of the tip. The second anode accelerates the electrons to the final energy determined by the voltage V between the cathode tip and the grounded second anode. A cold FEG is an excellent point source of illumination, and may not even require demagnification action from the first condenser lens. Another important advantage of cold FEG is the absence of thermal energy spread, so that the beam can be highly monochromatic. On the other hand, the emission current is destabilised by a contamination of the tip, and so regular maintenance (the so-called “flashing procedure”) is necessary. The electric field requirements and the demanding vacuum requirements of FEG can be reduced significantly by heating the tip. In a thermal FEG the emitter is heated to a temperature that is essentially lower than that of thermal emission. Owing to the reduction of the work function due to the strong electric field, the Schottky effect is the basis for electron emission. Usually, the work function of the emitter is also reduced by covering the tungsten tip with a thin layer of ZrO. The performance characteristics of a cold FEG are better than that of a thermal FEG, but the latter has the advantage of less emission noise and provides a stable emission current which is not influenced by contamination layers. A comparison of the characteristics of various electron guns is given in Table .. The performance characteristics of cold and thermal FEGs make them a good choice for modern analytical TEMs used in materials science, while a thermionic gun with a LaB emitter is a less expensive alternative for use in a modern TEM applied to investigate polymeric materials.

2.3 The Instrument

33

Table 2.4. Comparison of the characteristics of various electron guns Emission

Thermionic

Thermionic

Schottky

Field emission

Material Work function (eV) Working temperature (K) Emission current (Acm−2 ) Gun brightness (Acm−2 sr−1 ) Probe current range Crossover diameter (μm) Energy spread (eV) Vacuum requirements (Pa) Cathode lifetime (h)

W 4.5 2800 1–3 105 1 pA−1 μA 20–50 1–3 10−2 –10−3 100

LaB6 2.7 1400–2000 25–100 5  105 –5  106 1 pA−1 μA 10–20 0.5–2 10−3 –10−5 200–1000

ZrO/W 2.8 1400–1800 1000 108 1 pA−5 nA 0.1–1 0.3–1 10−7 –10−8 2000

W 4.5 300 105 –106 108 –109 1 pA−300 nA 0.01–0.1 0.2–0.5 10−8 –10−9 1000

Fig. 2.8. Arrangement of lenses and apertures in the illumination system, and illustration of the corresponding TEM (left) and spot (right) illumination for the microscope shown in Fig. 2.5 (according to the ray diagram in the manual of the LEO 912 OMEGA; reproduced with the permission of Carl Zeiss SMT AG)

2.3.2 Illumination System The illumination system transfers electrons from the gun crossover to the specimen under various illuminating conditions, ranging from a broad and essentially parallel beam to a focussed probe. In principle, a single, rather weak, condenser lens with a magnification of unity can project the electron crossover in the gun to a focussed spot in the specimen plane or to an unfocussed broader disc. However, the resulting focussed spot with a diameter of some tens of micrometres is much larger than is

34

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

needed to illuminate the specimen for high-magnification operation; for example, at a magnification of   the extension of the imaged specimen area is only about  μm. However, the irradiation of the specimen area should correspond as closely as possible to the viewing screen, whatever the magnification. This reduces specimen drift due to heating and limits the radiation damage and contamination in unirradiated areas. Therefore, a TEM has at least a double condenser system with a strong first lens to reduce the gun crossover to an image about  μm in diameter and a weaker second lens to project this image onto the specimen plane. Usually the number of lenses in the illumination system in modern microscopes is increased beyond this, however. This is because, on the one hand, a defined range of illumination apertures is required in the normal TEM mode, from around  mrad for medium magnifications to  . mrad for high-resolution and phase contrast microscopy to . mrad for small-angle electron diffraction and holographic experiments. On the other hand, modern TEMs are often also used for analytical investigations, and so with the aid of the illumination system a small electron probe appropriate for X-ray microanalysis, convergent beam electron diffraction techniques or scanning transmission mode must be produced. Therefore, a typical illumination system in a modern TEM consists of three condenser lenses and also includes the upper part of the objective as an objective prefield lens, and sometimes an additional condenser minilens is arranged between the third condenser lens and the objective prefield lens. As an example, Fig. . schematically illustrates the arrangement of lenses and the beam formation for TEM (left) and spot (right) illumination in the illumination system of the microscope shown in Fig. .. Condensers  and  together form a zoom lens system which images the crossover of the gun, which is variably reduced in the constant zoom image plane. Stray electrons are blocked from passing through the zoom system by a fixed aperture provided in the principal plane of condenser . For spot illumination, the third condenser images the crossover in the entrance image plane of the objective prefield lens, which produces crossover image on the specimen that is reduced by approximately , and the variable aperture placed in the principal plane of condenser  acts as condenser aperture for the focussed spot. In the TEM illumination mode, the third condenser transmits the crossover image of the zoom system to the front focal point of the objective prefield lens. The illumination of each object point is thus parallel with the electron-optical axis, and the diameter of the illuminated specimen area is determined by the variable aperture in the principal plane of the third condenser. In Fig. . a more detailed ray diagram of the section between condenser  and the specimen plane illustrates that the illuminating cones for each object point have the same illuminating aperture α i , given by the radius of the crossover image r co divided by the focal length f p f of the objective prefield lens. Therefore, the illumination aperture can be varied by changing the excitation of the zoom system consisting of condensers  and . It is necessary to have a convenient system for aligning the electron beam such that the latter passes the elements of the illumination system in a proper manner. The alignment steps needed are, on the one hand, the translation of the beam without altering its inclination and, on the other hand, the tilting of the beam such that it does

2.3 The Instrument

35

Fig. 2.9. Illustration of the illuminating cones for several object spots using corresponding ray diagrams (according to the ray diagram in the manual of LEO 912 OMEGA; reproduced with the permission of Carl Zeiss SMT AG)

not change the beam’s position in the reference plane. Both of these motions can be obtained by a pair of scan coils which cause appropriately balanced beam deflections in opposite directions. Such coils are also used to scan or rock the electron probe in the scanning mode or for special diffraction techniques, respectively, and to generate a dark-field TEM mode. 2.3.3 Objective Lens and Specimen Stage The combination of the objective lens and the specimen stage is of particular importance for the microscope. As already noted above, the aberrations of the objective lens determine the resolution in a fundamental manner, and so its astigmatism must be corrected by a stigmator placed behind the objective lens, and the region of lens field action must be as small as possible to reduce spherical aberration. The stage is used to clamp the specimen holder into the correct position such that the specimen is positioned in the pole piece gap with great adequacy and stability and the objective lens can form images and diffraction patterns in a reproducible manner. Additionally, the stage must provide orthogonal x and y movements in the specimen plane (typically  mm from its centred position on the microscope axis) to select specimen sites of interest for imaging. Furthermore, defined orientations of crystalline materials and tomographical investigations of specimens require a eucentric goniometric stage with tilt rotation or double tilt motion. There are goniometer stages that permit the specimen plane to be tilted by up to  , combined with rotation by  about the lens axis, or alternatively, simultaneous and independent tilts about two axes at right angles. The space between the pole pieces of the objective lens is however very limited, and particularly in TEMs equipped with special high-resolution pole pieces the narrowness of the gap in the objective lens limits the specimen tilt to significantly smaller values. There are two quite distinct types of specimen stages in

36

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

common use: top-entry stages and side-entry stages. The specimen holder in a topentry stage enters the bore of the objective lens from above. In a side-entry stage the specimen holder takes the form of a flattened rod and is introduced into the pole piece gap from the side. Top-entry stages are less sensitive to vibrations and thermal drift and are therefore used in a few HRTEMs, despite some drawbacks (limited tilting, unsuitability for analytical attachments, difficult to combine with airlocks and anticontamination). Modern side-entry stages have performances similar to those of the best top-entry stages and are the most common type used. They are easy to operate with an airlock and with a cryoshield around the specimen for reducing specimen contamination, and can also be well designed for analytical investigations and in situ manipulations (heating, cooling, straining) of specimens. Usually, polymeric ultrathin films for TEM investigations are fixed on a support consisting of a fine metal grid  mm in diameter (see Chap. ). The grid with the specimen can easily be transferred with the help of tweezers into a conventional side-entry specimen holder, where it is fixed in place by a special holding bow. The application of special holders for in situ heating, cooling and specimen straining are described in Chap. . 2.3.4 Image-Forming System The image-forming system of a TEM consists of at least three lenses: the objective lens, the intermediate lens and the projector lens. The function of the objective lens, to form an image of the illuminated specimen area, results from the postfield of the lens at the back side of the specimen, while the objective prefield lens is included in the illumination system, as explained above and illustrated in Fig. .. The intermediate lens can magnify the first intermediate image, which is formed just in front of this lens (Fig. .a), or the first diffraction pattern, which is formed in the focal plane of the objective postfield lens (Fig. .b) by reducing the excitation. One of these images produced by the intermediate lens is enlarged by the subsequent projector lens and displayed on the viewing screen. The bright-field mode (Fig. .a) with a centred objective aperture in the back focal plane of the lens is the typical TEM mode, and the scattering contrast caused by this aperture (see Sects. .. and .) is widely used for TEM investigations of polymers. In the so-called selected area diffraction mode (Fig. .b), a diffraction pattern of a reduced specimen area is usually projected onto the viewing screen. This is carried out by selecting the area of interest via a corresponding aperture (SAD aperture) in the plane of the first intermediate image formed by the objective lens. Real imaging lens systems in TEM instruments are far more complicated than the simple three-lens model of Fig. .. An increased number of lenses and additional alignment elements tuned to each other with the aid of computer control are necessary to change the magnification across a wide range from about  up to    in an appropriate manner, to compensate for image rotations and to provide different imaging modes. To understand the optics of any particular microscope, it is necessary to consult the ray diagrams in the manufacturer’s operating manual.

2.3 The Instrument

37

Fig. 2.10a,b. Two basic operation modes of the TEM image-forming system: a bright-field image mode, b selected area electron diffraction mode

2.3.5 Viewing Chamber and Image Acquisition The final electron microscopic image is projected onto a luminescent screen in the viewing chamber, which can be viewed via a window made of lead glass. An attached pair of binoculars is usually used to observe details in the image. For a long time the recording of images in a TEM was based on the exposure of electronsensitive negative films. This has changed recently with development of slow-scan CCD cameras and image plates. Characteristic features of the different recording techniques are briefly described below. Negative Film Traditionally, electron-sensitive photographic emulsion layers on polymer (formerly glass) support bases have been used for image recording. The sensitivity of the film to

38

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

electrons is similar to their sensitivity to light. The camera system is placed within the microscope immediately below the viewing chamber. A light-tight dispensing magazine which contains about  cassettes with film sheets of size, e.g.,  mm   mm is loaded into the camera chamber. In order to record an image, one of the cassettes is transferred from the light-tight box into the position below the screen where the electron image falls on it when the screen is tilted out of the electron beam, and an electrically operated shutter is opened for a preset exposure time. After exposure, the film sheet is transferred into the same or another storage box, from where it can be retrieved later for processing to a permanent negative image. Although film is the oldest recording medium, it still retains sufficient advantages to find use in current systems. Compared to digital recording media, film has significantly better resolution, which depends on the type of the film and is around  μm in most cases. With its high resolution and large area, film gives the largest field of view. Film is also an excellent archiving media, and we know from experience that the original image information can be stored in film for a long time. Unlike film, digital images are stored with today’s technologies, including data formats, storage media and reading software, and hardware devices. Ensuring the survival of these images is by no means a simple task, and this is still an unsolved challenge. A quantitative analysis of an image stored in a film can be done with the aid of a film scanner, by measuring the optical density (which, however, falls within a small range from . to . eμm− ), which is proportional to exposure. The total dynamic range is also rather limited (about  ). These properties permit normal imaging applications, but limit the application of film to low-dose imaging and for recording images with very strong contrast, as typically observed for diffraction patterns. Slow-Scan Charge-Coupled Device (SSCCD) Cameras SSCCD cameras have high sensitivity, a linear response over a wide range of intensities (four orders of magnitude), and offer a direct path to digital data acquisition. However, with a pixel size of between  μm and  μm and typically    or    pixels, a much smaller image area is covered than in image recording with negative films. In principal, incident high-energetic electrons could be directly detected by a CCD, but the direct exposure of the CCD to the electron beam causes radiation damage, and due to the high number of thermalised electrons caused by each of the high-energetic electrons, the CCD will be saturated by a relatively small number of incident electrons. Therefore, current SSCCD cameras used in a TEM record an electron image in three stages: () incident electrons are converted into light by an yttrium-aluminium-garnet (YAG) single-crystal scintillator or a powder phosphor; () released photons are transported to the CCD array via a fibre optic plate or lens couplings; () the light is converted to an electron charge that is temporarily stored in each channel of the CCD, where the CCD is an array of metal-oxidesemiconductor (MOS) capacitors. By controlling the gates of the MOS capacitors, charge can be accumulated and transferred. For the SSCCD, the charge is transferred row by row to the serial register placed next to the CCD array, and then transferred pixel by pixel and measured by an analogue/digital converter. Usually a SSCCD camera is cooled by liquid nitrogen or by an electronic Peltier cooler in order to reduce

2.3 The Instrument

39

the dark current (which produces noise). Images can be acquired, viewed and processed immediately with a SSCCD camera. Thus the quality of images can be assessed during the operation of the microscope, which is very useful when image recording depends critically on the microscope operation conditions. Image Plate (IP) The image plate used in standard film cassettes is a reusable flexible sheet. It consists of a layer of storage phosphor, where tiny crystals of (typically) BaFBr:Eu+ are embedded in a resin. This active layer is coated onto a polyester support sheet and covered by a polymeric protective layer. The electron irradiation results in Eu+ Eu+ transitions and excites the crystals at their luminescence centres to a semi-stable state. The image information created by this excitation is stable for many hours and decays within days. By irradiating them with red laser light with a wavelength of  nm, the crystals are stimulated to release the stored information as a blue luminescence signal with a wavelength of  nm. The storage process itself is highly localised, but due to the lateral scattering of the incident electrons within the active layer of the IP, it is limited to a few μm. Finally, however, the spatial resolution of an IP is determined by the read-out process. The IP must be transferred to a special external read-out device, where it is scanned and the stored information is read out by a moving unit containing the laser and a detection system for the luminescence signal released. Depending on the instrument, the read-out pixel sizes of an IP with an effective area of  mm   mm are between  μm and  μm. The luminescence signal emitted from the plate while reading is directly proportional to the exciting electron dose, and the dynamic range is more than six orders of magnitude. 2.3.6 Alignment and Operation of Transmission Electron Microscopes No TEM will give its best performance unless its electron-optical elements are aligned with each other, the astigmatism of the illumination system and the objective lens is corrected by corresponding stigmators, and the beam-limiting diaphragms are accurately centred on the electron-optical axis. The lenses and other components of the column of a modern electron microscope are accurately aligned mechanically by the manufacturer. However, there is always the need for minor adjustments, which are performed by the operator using sets of double deflection coils placed at strategic points in the column. Alignment procedures differ from microscope to microscope, but in principle they are based on the following steps []. First, without the presence of a specimen, the emission current of the electron gun must be adjusted. The focussed beam is projected onto the screen, and under unsaturated conditions (corresponding to certain settings of the filament temperature and bias voltage) a symmetrical filament pattern is formed with the aid of gun tilt coils. Then the illumination system must be aligned. This includes correctly positioning the condenser aperture and correcting for astigmatism using the condenser stigmator. Typically, a condenser alignment procedure centres the focussed beam at different lens currents of the first condenser lens (variation of spot size), and this procedure is repeated until the beam is found to occur at the centre of the screen (no shift) irrespective of the spot size.

40

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

After the alignment of the condenser lens system the specimen is placed in the reference plane, where the image on the screen is in focus at the optimum value of the objective lens current. This reference plane is also the eucentric plane of a side-entry holder; this means that a point on the electron-optical axis does not move laterally when tilted around the holder axis. An image wobbler is usually used to find the exact position of the reference plane by varying the z-control of the sample holder. This produces a superposition of two images on the screen as long as the specimen is placed below or above the reference plane, and the distance between the corresponding image structures in the double image increases with increasing distance between specimen plane and reference plane. With the specimen in the reference plane, the rotational centre of the objective lens can be checked. The illumination is tilted if wobbling the strength of the objective lens results in an image shift. The adjustment of the illumination tilt is however often undertaken by varying the beam voltage with a wobbler (adjustment of the voltage centre), as recent instruments show little instability in the objective lens current, so image broadening is mainly caused by the change in wavelength due to the inelastic scattering of electrons in the specimen. The astigmatism of the objective lens is adjusted with the objective lens aperture inserted. As the astigmatism depends on the aperture size and its position, the chosen aperture is inserted and centred in the selected area diffraction mode, and then, after switching back to the normal image mode, the lens and aperture can be stigmated as a unit. To correct the astigmatism at magnification factors of up to several hundreds of thousands, the features of the Fresnel fringes at thin edges or small holes of an amorphous specimen (e.g. carbon) are usually observed at different focus conditions. For high-resolution imaging at magnifications of  , the streaking in the image is used for astigmatism adjustment. Other alignments in subsequent parts of the image forming system are less crucial and are less likely to change from day to day. If, for example, a feature of interest located at the centre of the viewing screen does not remain there or even moves out of the viewing area as the magnification is varied, this is a typical indication of the misalignment of the intermediate or projector lenses. Correction is achieved using a set of one or more deflection coils for image shifts below the objective lens. This misalignment predominantly appears at magnification steps where continuous changes in magnification can only be achieved by changing the lens currents of some lens via a computer. These magnification changes are construction-dependent features of the particular microscope used, and so such adjustments must be carried out according to special routines described in the microscope’s manual.

2.4 Fundamentals of Image Formation Image contrast is caused by the different scattering mechanisms that occur in amorphous and crystalline materials. However, the intensity distribution of the image depends not only on the interaction of the electron beam with the object, but also on the illumination conditions and in particular on the action of the objective lens and the arranged apertures.

2.4 Fundamentals of Image Formation

41

2.4.1 Scattering Mechanism and Contrast Formation In general, electron scattering (see, e.g. [,–]) can be divided into two types: inelastic and elastic scattering. Inelastic scattering is an electron–electron interaction where electrons scattered by the atomic shell electrons suffer a considerable loss of energy. Elastic scattering can be considered to be an electron–nucleus interaction in which the atomic electron cloud merely screens the Coulomb potential. However, for small scattering angles θ, a negligible amount of energy is transferred to the nucleus and so the electrons are scattered with no appreciable energy loss. This small-angle elastic scattering is the most important of all interactions in polymeric materials that cause contrast in the electron image. For larger scattering angles, and also for high electron energies, an appreciable energy transfer from the electron to the nucleus results from the elastic electron–nucleus interaction process. These electron–nucleus collisions can lead to specimen damage if the energy transfer exceeds the threshold energy (on the order of – eV), such that an atom can be displaced from its original position in the crystal to an interstitial site. A knowledge of the angular distribution of the elastically scattered electrons is the basis for contrast interpretation. Elastic electron scattering is illustrated in Fig. . in the particle model (a) and in the wave model (b). Electrons that pass through an element of area dσ of the parallel incident beam will be scattered via Coulomb forces into a cone of solid angle dΩ in the particle model. The inclination of the scattered electrons is expressed by the scattering angle θ. In the wave model, the incident electron beam is described by a (time-invariant) plane wave ψ = ψ  eπ i k  z

(.)

with an amplitude ψ  and a magnitude k of the wave vector whose direction coincides with the coordinate z. Using a spherical coordinate system with its origin at the scattering centre, the scattering event is described by a spherical scattered wave  ψ sc = ψ  f (θ) eπ i kr r

(.)

with a magnitude k of the wave vector and an amplitude that depends on the scattering angle θ. Far from the nucleus, the total wave field is expressed as the superposition of the undisturbed plane incident wave and the spherical scattered wave  ψ s = ψ  eπ i k  z + iψ  f (θ) eπ i kr r

(.)

where the phase shift of  between the scattered and the incident wave is taken into account by the factor i of the second item. In general, the scattering amplitude f (θ) is a complex quantity and can be expressed by f (θ) =  f (θ)e i η(θ) .

(.)

This means that there is not only the mentioned phase shift of  but also an additional phase shift η(θ), which depends on the scattering angle.

42

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation Fig. 2.11. Elastic electron scattering in the particle model (a) and scattering in the wave model with the superposition of a plane incident wave and a spherical scattered wave (b)

The most convenient quantity for characterising the angular distribution of scattered particles is the differential cross-section: the ratio dσ dΩ. This cannot be calculated exactly from the classical particle model, but using the wave model the relation dσel =  f (θ) dΩ

(.)

between the elastic differential cross-section dσel  dΩ and the scattering amplitude f (θ) of a single atom can be derived. The scattering amplitude f (θ) can be calculated from the wave-mechanical Schrödinger equation. However, a detailed treatment of this problem is beyond the scope of this book, and the reader is referred to textbooks on transmission electron microscopy (see e.g. []), where theoretical and experimental results are discussed. There is a simple geometric relationship between the scattering angle θ and the solid angle Ω Ω = π( − cos θ)

(.)

dΩ = π sin θ dθ

(.)

and therefore

is obtained when Eq. . is differentiated. Using Eq. ., the following differential relation for dσel can be written: dσel =

dσel dσel dΩ dθ = π sin θ dθ . dΩ dθ dΩ

(.)

2.4 Fundamentals of Image Formation

43

By integrating Eq. . over θ from α to π, the number of electrons elastically scattered through angles θ  α can be calculated. This gives the partial elastic crosssection σel (α): π

∫

σel (α) =

α

dσel π sin θ dθ . dΩ

(.)

The total elastic cross-section σel is obtained if the lower limit of integration is zero, i.e., π

σel =

∫



dσel π sin θ dθ . dΩ

(.)

Taking into account the corresponding expression for inelastic scattering, the total scattering cross-section σt of an atom is given by σt = σel + σinel ,

(.)

where σinel denotes the total inelastic cross-section. The total scattering cross-section σ t determines whether or not scattering occurs and can be regarded as the effective target area presented by the scatterer. The description of scattering by extended specimens must take into account possible mutual interference of the electron waves scattered at the atoms. Therefore, the contributions of the individual scatterers to the intensity in the image plane depend on the atomic arrangement, and so amorphous objects and crystalline objects – the two limiting cases of disorder and order, respectively – contribute in different ways to the interaction events. For an (ideal) amorphous object, the electrons are scattered incoherently, i.e. systematic correlations of the phases do not have to be taken into account, and in order to determine the intensity in the final image plane the intensities and not the wave amplitudes of the individual scattering events are summed. Assuming that the specimen consists of several layers of thickness dx where the scatterers do not overlap, there will be NA ρ dx (.) A atoms per unit area in each layer. Here N A is Avogadro’s number, A is the atomic weight and ρ is the density of the specimen. As the N scatterers block an area equal to N σt , the intensity I of the incident electron beam is lowered by an amount dI in each layer, and this can be expressed as N=

dI NA =− σt ρ dx (.) I A The integration of Eq. . over a finite thickness x yields the following relation for the intensity I as decreased by scattering I = I e−

NA σ ρx A t

,

(.)

where I denotes the intensity of the incident beam. In the bright-field mode, as described in Sect. .., the diaphragm in the focal plane of the objective lens acts as

44

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

a stop that absorbs all electrons scattered through angles θ  α, where α is determined by the aperture hole. To obtain the decrease in transmission for this case, the total cross-section σ t in Eq. . must be replaced by the sum of the partial crosssections σel (α) and σinel (α) for elastic and inelastic scattering, respectively (compare Eq. .). Scattering contrast is caused by the local density ρ and the local specimen thickness x in the same way, and so the product of the density and the thickness is often referred to as the mass thickness and the corresponding contrast is called the mass-thickness contrast. Although in practice an ideal amorphous object with complete random disorder does not exist, this contrast describes the image intensity for low and medium magnifications well, unless a highly coherent electron beam and large defocusing are employed. However, at high magnifications and electron-optical imaging conditions, which permit the resolution of details that are smaller than  nm, a contrast interpretation must take into account that according to the wave-optical theory of imaging it is necessary to sum over the (complex) wave amplitudes and obtain the image intensity by squaring the wave amplitude in the final image plane. This results in a prevailing phase contrast of a thin amorphous weak scattering object, as described in more detail in Sect. ... 2.4.2 Electron Diffraction and Diffraction Contrast The electron beam interacts with the specimen in a very specific way when the specimen is crystalline. In crystalline structures, the constituent atoms are geometrically arranged in a regular pattern, based on a unit cell which is repeated in all three dimensions throughout the crystal. Normal to well-defined directions within the lattice there are sets of equidistant, parallel lattice planes. Each set of corresponding lattice planes is characterised by a triple of Miller indices h, k, l and an interplanar distance d hk l between neighbouring lattice planes. Unlike the incoherent scattering in amorphous specimens, in a crystalline object the incident electrons are scattered coherently in well-defined directions at the atomic arrays. This scattering phenomenon is called diffraction, and the interference maxima occur according to the Bragg law: nλ = d hk l sin Θ .

(.)

Here λ is the wavelength of the incident electrons, n is an integer and determines the order of diffraction, and Θ is the Bragg angle, which is half of the scattering angle θ, i.e. θ = Θ. Relative to the lattice planes, the angle of incidence and the Bragg angle Θ are equal. This means that diffraction in crystalline materials, which is strictly an interference phenomenon, can be interpreted as a reflection of the incident electrons at the lattice planes. As the wavelength λ of the electrons used in a TEM is about two orders of magnitude lower than typical values of d hk l , the corresponding values of Θ are very small. This means that, in practice, an electron beam will only be strongly diffracted from lattice planes which are almost parallel to the electron beam. On the one hand, this factor makes the geometry of electron diffraction patterns much simpler than that of X-ray diffraction patterns, for which Θ can be very large. On the other hand, the accuracy of a structural analysis performed via electron diffraction is rather limited by the relatively small values of the Bragg angles. As described in

2.4 Fundamentals of Image Formation

45

Sect. .., the diffraction pattern is formed in the focal plane of the objective postfield lens, where each diffracted beam is focussed into a corresponding spot according to its appearance as an oblique parallel beam. With the aid of the subsequent intermediate lens and the projector lens, this diffraction pattern is enlarged and displayed on the viewing screen. Usually, the selected area diffraction mode (Fig. .b) is applied to display a diffraction pattern of the reduced specimen area of interest. When the TEM is used in the image mode, some of the diffracted electrons can be removed from the image plane by the objective aperture. In the bright-field image this aperture is used to stop all diffracted electrons and only undiffracted electrons are allowed to contribute to the image. A bright-field image of a perfect, unbent crystal of uniform thickness does not produce any contrast. Contrast is formed if the diffraction conditions for the electrons vary locally, e.g., due to variations in the sample thickness, as a result of bending the crystal, or due to the presence of crystal defects. It is often more informative to use diffracted electrons rather than the transmitted beam, since these electrons have actually interacted with the specimen and image contrast is usually also enhanced. There are two ways to produce these dark-field images. The first and simplest way is to displace the objective aperture sideways so that the image is formed with a corresponding off-axis diffracted beam. This easy mode of operation has the disadvantage that off-axis rays are used for imaging and image quality is adversely affected by aberrations in the lens. Therefore, in the alternative second mode the objective aperture is kept centred on the electron-optical axis, and the incident beam is moved off-axis by tilting the illumination. The main interest in diffraction contrast arises from the ability to make crystal defects (such as dislocations, stacking faults, grain boundaries and precipitates) visible. The kinematical or the dynamical theory of electron diffraction must be applied in order to understand the contrast that arises from defects. Often a contrast interpretation based on the two-beam approximation can be used, where only the primary beam and one diffracted beam are strongly excited. The kinematical theory assumes that the amplitude of a wave diffracted according to the Bragg law is small compared to that of the primary wave. This approach is only applicable to very thin specimens. The dynamical theory, based on the Schrödinger equation, also describes wave propagation in thick crystals. In the periodic potential of a crystal lattice, the electron waves propagate as a Bloch-wave field, which exhibits the same periodicity as the lattice. With the aid of the dynamical theory, the Bloch-waves can also be calculated for the so-called n-beam case, which considers the transmitted beam and n− diffracted beams and is used for high-resolution imaging of crystal lattice structures. Visualising defects by diffraction contrast and high-resolution imaging modes are very important applications of TEM investigations in the field of inorganic materials (see, e.g.,[, –]). However, due to the electron beam sensitivity of polymers, diffraction contrast does not play a major role when studying crystalline polymers by means of TEM. Therefore, a more detailed description of the contrast formation, the special investigation techniques and the theoretical models used to interpret experimental results are beyond the scope of this book.

46

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

2.4.3 Fundamentals of the Imaging Process The image contrast obtained is mainly determined by two processes: the electron scattering in the sample (Sects. .. and ..) and the electron-optical imaging process, including the microscope aberrations. The latter can be described in detail on the basis of the wave-mechanical theory of contrast formation (see, e.g., [, , , –]). As illustrated in Fig. ., the overall process of image formation can be roughly explained as follows. The interaction of the incident electron beam with the specimen results in an object wavefunction ψob (x) at the exit surface of the specimen (object plane). Here and in the following, for sake of a simplified notation, only the functions for a one-dimensional object are taken into consideration. Far away from the object, the scattered electron wave forms a Fraunhofer diffraction pattern, which can be observed using the objective lens in its back-focal plane. This step is mathematically described by the Fourier transform of the object wavefunction ψob (x), which leads to the wavefunction Ψd (u), where u is the spatial frequency (i.e. the coordinate in the reciprocal space), which corresponds to the scattering angle θ according to the relation θ = uλ .

(.)

The propagation of the wave from the back-focal plane to the image plane can be described by the inverse Fourier transform, if for simplicity the magnification factor is not taken into consideration. Thus, the action of an ideal lens without imperfections would result in an image wavefunction ψ i (x), which directly corresponds to ψob (x). However, taking into account real imaging conditions, the effect of the objective lens on the propagating wave can be expressed by multiplying the wavefunction Ψd (u) in the back-focal plane of the lens by the transfer function T(u), which is composed of three multiplicative parts: T(u) = A(u)C(u)D(u) .

(.)

The value of the real aperture function A(u) is directly related to the geometrical position and size of the diaphragm, i.e. A(u) =  for the open parts of the aperture, and A(u) =  for the opaque parts. The specimen-independent properties of the electron-optical imaging system are characterised by the contrast transfer function C(u). In the case of perfectly coherent axial illumination, C(u) describes the phase shift of the electron wave due to the influence of the spherical aberration and the defocussing of the objective lens by the following equation: C(u) = e i χ(u ,Δ f ) = e  (C s λ πi



u  −λΔ f u  )

.

(.)

Here, Cs denotes the spherical aberration coefficient, Δ f the defocus and λ the wavelength of the incident electrons. In reality, electron sources are not coherent; their finite chromatic aberrations and beam divergences result in a damping of the high-resolution information. In general,

2.4 Fundamentals of Image Formation

47

Fig. 2.12. Schematic representation of the Abbe formulation of the imaging process

envelope functions are a useful means for describing the effect of damping. Thus the damping function D(u) can be written as D(u) = D (u)D (α i , u) .

(.)

Here, D is the damping envelope due to focus spread and D is the damping due to beam divergence, as denoted by the illumination aperture α i . In practice it is usually sufficient to take into consideration a certain spread of defocus δ f . With this standard deviation δ f in the range of – nm, the envelope function D is given by D (u) = e−  π 

    λ δf u

.

(.)

D (u) damps high spatial frequencies significantly and limits the spatial resolution attainable. In order to calculate the image wavefunction ψ i (x) and the image intensity I(x) given by 

I(x) = ψ i (x) ,

(.)

it is useful to take advantage of the spread function t(x), which is the Fourier transform of the transfer function T(u). Since the Fourier transform of the product of

48

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

two functions is the convolution of their Fourier transforms, ψ i (x) in Eq. . can be replaced by the convolution of the functions ψob (x) and t(x), so that the image intensity becomes I(x) = ψob (x)  t(x)



(.)

where  is the convolution operator. A relatively simple relationship which is better at showing the influence of the transfer function can be derived if a very thin specimen is taken into consideration. This acts as a pure phase object, and the object wavefunction ψob (x) at the exit surface of the specimen is related to the incident wave ψ  as follows: ψob (x) = ψ  e−i σΦ(x) .

(.)

Here, Φ(x) is the projection in the beam direction of the electrostatic potential distribution in the specimen and σ is the interaction constant given by σ=

π λV

(.)

where V is the accelerating voltage. If the amount of phase shift is small, the exponential function in Eq. . may be expanded and higher order terms neglected. This results in a normalised function ψob,n (x) with ψob,n (x) =

ψob (x) =  − iσΦ(x) ψ

(.)

which is known as the weak phase object approximation (WPOA). Taking into account that the complex spread function t(x) is composed of a real part denoted by c(x) and an imaginary part denoted by s(x), Eq. . reduces to I(x) =  + σΦ(x)  s(x)

(.)

if ψob (x) is replaced by the function ψob,n (x) and terms of second order in σΦ(x) are omitted. Thus, the image intensity directly represents the projected potential of the sample, smeared out by convolution with the imaginary part s(x) of the spread function t(x), which therefore determines the resolution in the image. The optimum resolution is given when s(x) is a sharp positive or negative peak []. As s(x) is the Fourier transform of the imaginary part of the transfer function T(u), this will occur when sin χ(u) is equal to + or − over as large a region of the reciprocal space as possible. Owing to the dependence of χ(u) on the defocus Δ f (see Eq. .), the overall form of sin χ(u) and therefore the passband of spatial frequencies which contribute to the image are controlled by focus-setting. Using the parameters V =  kV (i.e. λ = . pm) and Cs = . mm, the defocus dependence of the contrast transfer function is demonstrated in Fig. . for defocus values of − nm (a), − nm (b) and − nm (c).

2.4 Fundamentals of Image Formation

49

Fig. 2.13. The functions sin χ(u) and cos χ(u) for 120 keV electrons with Cs = 2.7 mm for defocus values of 30 nm (a), 110 nm, i.e. Scherzer focus (b), and 1000 nm (c)

50

2 Transmission Electron Microscopy: Fundamentals of Methods and Instrumentation

Scherzer was the first to describe the focus setting for optimum phase contrast transfer []. Plot (b) in Fig. . represents the contrast transfer function at the socalled Scherzer focus which, in accordance with [], is given by 

  Δ f = −  Cs λ . 

(.)

At this slight underfocus of the objective lens, sin χ(u) decreases to − and is nearly equal to − for a relatively broad range of spatial frequencies u before increasing to zero at u z with u z = . Cs λ  

− 

.

(.)

Since for higher values of u the signs of the contributions to the image are reversed and than oscillate rapidly, an objective aperture must be inserted to cut off all values of u that exceed u z . In practice, this is often not necessary, as high spatial frequencies are significantly damped according to the damping envelopes of D(u). The reciprocal Δ x of the first zero u z defines the “interpretable resolution” under optimum conditions, which corresponds to “point resolution” in the case of nonperiodic objects. At the Scherzer focus condition, this resolution is given by 

Δ x = . Cs λ    .

(.)

References 1. Haguenau F, Hawkes PW, Hutchison JL, Satiat-Jeunemaître B, Simon GT, Williams DB () Microsc Microanal : 2. Mulvey T () Brit J Appl Phys : 3. Cosslett VE () J Microsc : 4. Ruska E () The early development of electron lenses and electron microscopy (Trans Mulvey T). S Hirzel Verlag, Stuttgart 5. Bethge H () Ultramicroscopy : 6. Hawkes PW (ed) () The beginnings of electron microscopy. In: Advances in electronics and electron physics. Academic, New York 7. Hashimoto H () J Electron Microsc Tech : 8. Marton L () Early history of the electron microscope, nd edn. San Francisco Press, San Francisco, CA 9. Phillipp F, Höschen R, Osaki M, Möbius G, Rühle M () Ultramicroscopy : 10. Midley PA, Weyland M () Ultramicroscopy : 11. Leapman RD, Kocsis E, Zhang G, Talbot TL, Laquerriere P () Ultramicroscopy : 12. Thomas JM, Midley PA, Yates TJV, Barnard JS, Raja R, Arslan I, Weyland M () Angew Chem : 13. Weyland M, Midgley PA () Materials Today : 14. Sugimori H, Nishi T, Jinnai H () Macromolecules : 15. Kübel C, Voigt A, Schoenmakers R, Otten M, Su D, Lee T-C, Carlsson A, Bradley J () Microsc Microanal : 16. Bals S, Van Tendeloo G, Kisielowski C () Adv Mater : 17. Hayder M, Uhlemann S, Schwan E, Rose H, Kabius B, Urban K () Nature :  18. Lentzen M, Jahnen B, Jia CL, Thust A, Tillmann K, Urban K () Ultramicroscopy : 19. Batson PE, Dellby N, Krivanek OL () Nature :

References

20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46.

47. 48.

51

Batson PE () Ultramicroscopy : Krivanek OL, Nellist PD, Dellby N, Murfitt MF, Szilagyi Z () Ultramicroscopy : Bleloch A, Lupini A () Materials Today : Hetherington C () Materials Today : Freitag B, Kujawa S, Mul PM, Ringnalda J, Tiemeijer PC () Ultramicoscopy : van der Stam M, Stekelenburg M, Freitag B, Hubert D, Ringnalda J () Microsc Anal :(EU) Rose H () Nucl Instrum Methods Phys Res A : Glaser W () Grundlagen der Elektronenoptik. Springer, Wien El-Kareh AB, El-Kareh JCJ () Electron beams, lenses, and optics. Academic, New York Grivet P () Electron optics, nd edn. Pergamon, Oxford Hawkes PW () Electron optics and electron microscopy. Taylor & Francis, London Hawkes PW, Kasper E () Principles of electron optics, vol : Basic geometrical optics; vol : Applied geometrical optics. Academic, London Reimer L () Transmission electron microscopy: physics of image formation and microanalysis, rd edn. Springer, Berlin Sawyer LC, Grubb DT () Polymer microscopy. Chapman and Hall, New York Shindo D, Oikawa T () Analytical electron microscopy for materials science. Springer, Tokyo Chescoe D, Goodhew PJ () The operation of transmission and scanning electron microscopes (Royal Microscopical Society Microscopy Handbooks ). Oxford University Press, New York Buseck PR, Cowley JM, Eyring L (eds)() High-resolution transmission electron microscopy and associated techniques. Oxford University Press, Oxford Fultz B, Howe J () Transmission electron microscopy and diffractometry of material, nd edn. Springer, Berlin Ernst F, Rühle M (eds)() High-resolution imaging and spectrometry of materials. Springer, Berlin Bethge H, Heydenreich J (eds)() Electron microscopy in solid state physics (Materials Science Monographs ). Elsevier, Amsterdam Spence JCH () Experimental high-resolution electron microscopy, nd edn. Oxford University Press, New York Williams DB, Carter CB () Transmission electron microscopy: a textbook for materials science. Plenum, New York Shindo D, Hiraga K () High-resolution electron microscopy for materials science. Springer, Tokyo Goodhew PJ, Humphreys FJ, Beanland R () Electron microscopy and analysis, rd edn. Taylor & Francis, London Zhang X-F, Zhang Z (eds)() Progress in transmission electron microscopy : Concepts and techniques; : Applications in materials science. Springer, Berlin Lenz F () Transfer of image formation in the electron microscope. In: Valdré U (ed) Electron microscopy in materials science. Academic, New York, p  Erickson HP () The Fourier transform of an electron micrograph: First order and second order theory of image formation. In: Barer R, Cosslett V (eds) Advances in optical and electron microscopy. Academic, London, p  Coley JM () Diffraction physics, nd rev edn. North-Holland, Amsterdam Scherzer O () J Appl Phys :

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

Studies of polymeric materials using transmission electron microscopy are often directed towards observations of the morphologies of these materials at low-tomedium magnifications. Therefore, in practice most investigations use conventional transmission electron microscopy and take advantage of the mass-thickness contrast of selectively stained polymer samples. Nevertheless, a lot of investigations of polymers have also been carried out by applying special investigation techniques. This chapter reviews these special methods and techniques, including electron diffraction, high-resolution transmission electron microscopy, phase contrast transmission electron microscopy, electron holography, low-voltage transmission electron microscopy, high-voltage transmission electron microscopy, electron tomography, analytical transmission electron microscopy and scanning transmission electron microscopy, as well as their application to polymeric samples.

3.1 Conventional Investigations Utilising Mass-Thickness Contrast The main purpose of performing studies using TEM is to observe the morphologies of polymeric materials at low-to-medium magnifications. Therefore, in practice most TEM investigations do not use special techniques and make use of mass-thickness contrast for contrast formation. Many examples of this approach are given in this book in Chaps. –, which deal with investigations of different kinds of polymeric materials. Figure . illustrates the two typical features of mass-thickness contrast by showing a TEM micrograph of a test specimen with both small gold particles and much larger polystyrene (PS) spheres distributed on a thin carbon support film. Apart from the intensity modulations within the particles due to diffraction contrast, the small gold particles appear dark owing to the high density of gold. The influence of a locally increased specimen thickness on contrast formation is shown by the PS particle in the centre of the image. Even though the thickness of the PS particle is much larger than that of the gold particles, as schematically illustrated by the side view of the specimen, it does not appear darker than the gold particles, as the density of PS is about twenty times lower than that of gold. Generally, polymeric materials are composed of only low atomic number elements with similar density. Therefore, significant contrast due to inherent variations

54

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

Fig. 3.1. Illustration of mass-thickness contrast: TEM micrograph of a test specimen with small gold particles and polystyrene latex spheres distributed on a thin carbon support (top) and schematic side view of the specimen (bottom)

of the local density cannot be expected in an image focussed in the Gaussian image plane. Mass-thickness contrast in chemically untreated polymers can be caused by a locally varying specimen thickness. On the one hand, it is known that contrast appears between different parts of heterogeneous polymers under the action of the electron beam due to beam damage, if the thickness is reduced in one part by the evaporation of volatile fractions of polymer chains []. On the other hand, so-called strain-induced contrast can result in a thin section of originally homogeneous thickness if the sample is stretched in a tensile stage []. Areas which are differently strained will be made visible by a correspondingly different change in thickness; typical structures are bright-appearing local shear bands, homogeneous or fibrillated crazes, or even voids in the vicinity of modifier particles. Mass-thickness contrast due to locally varying specimen thickness is also the basis for imaging the surface relief of bulk materials using replicas. Often the replication technique is combined with metal shadowing to enhance the contrast. Shadowing is formed by coating the specimen surface with a thin metal layer (typically consisting of platinum–palladium) such that the metal vapour meets the surface at oblique incidence, so that the metal is preferentially condensed on the high points of the sample surface (see Sect. .). Selective staining is a very useful preparation technique that makes the morphology of polymer samples visible by enhancing mass-thickness contrast in TEM micro-

3.2 Electron Diffraction

55

graphs. It takes advantage of the different depositions of heavy-metal compounds, e.g. RuO or OsO , in the amorphous and crystalline regions of a polymer or of the different affinities of the stain due to the chemical and physical natures of the components of the polymer. These preparation techniques are described in more detail in Chap. .

3.2 Electron Diffraction 3.2.1 Selected Area Diffraction As described in Sect. .., the diffraction pattern is formed in the focal plane of the objective postfield lens, where each diffracted beam is focussed into a corresponding spot according to its appearance as an oblique parallel beam. With the aid of the subsequent intermediate lens and the projector lens, this diffraction pattern is enlarged and displayed on the viewing screen. Usually, the selected area diffraction mode (Fig. .b) is applied to display a diffraction pattern of a reduced specimen area of interest. Figure . shows typical sets of TEM micrographs and corresponding diffraction patterns for changes in the supramolecular structure in fractions of linear polyethylene (PE) with increasing molecular weight. While highly oriented lamellae produce a diffraction pattern composed of individual spots, irregularly arranged crystalline lamellae correspond to a polycrystalline material and cause a ring pattern in the diffraction plane. The ring pattern becomes arced with increasing orientation of the lamellae. Therefore, electron diffraction is sometimes used as a means to evaluate the orientation of the structural entities (lamellae, fibrils, microfibrils) of a polymer. 3.2.2 Structure Analysis Direct structure analysis of small crystals is possible when a goniometer stage is also used, since it allows three-dimensional sampling of the crystal through the use of specimen tilt. Therefore, electron diffraction seems to be a very useful technique for studying crystal structures of small dimensions in polymers, e.g. crystalline lamellae of semicrystalline polymers. However, radiation damage is the main limitation on those applications and a diffraction pattern of a polymeric single crystal usually disappears in a very short time under conventional image conditions. Nevertheless, a few research groups have successfully applied electron diffraction analysis in order to study polymer crystal structures. Recent reviews of these special investigations of polymers via TEM have been given by Dorset [–] and Voigt-Martin []. However, it should be mentioned that during diffraction work on sensitive samples the experimentalist must do everything to reduce radiation damage, for example by cooling the specimen down to very low temperatures and working at very low electron illumination levels (see Sect. .).

56

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

Fig. 3.2a–d. Changes in the supramolecular structure in fractions of high-density polyethylene (HDPE) with increasing molecular weight MW as revealed by HVTEM bright-field images of thin sections cut by means of a cryoultramicrotome (left column), TEM bright-field images of stained ultrathin sections (middle column), and diffraction pattern (right column); a MW = 8700: sheaflike structure, formed by long, parallel lamellae; b MW = 34 000: bundle of long lamellae; small variations in local orientation; c MW = 143 000: spherulites with concentric rings; rings are constructed from short lamellae which periodically change their orientations; minor local orientation; d MW = 579 000: very short lamellae are irregularly arranged; no preferred orientation (from [60], reproduced with the permission of Carl Hanser Verlag)

3.3 High-Resolution Transmission Electron Microscopy

57

3.3 High-Resolution Transmission Electron Microscopy 3.3.1 Introduction While a resolution of below a couple of nanometres should not be expected for conventional TEM investigations of polymers, high-resolution transmission electron microscopy (HRTEM) corresponds to imaging at a resolution sufficient to resolve, for example, the local packing of molecules into a crystalline lattice. On the one hand, this requires a TEM with adequate equipment and perfect beam alignment and, on the other, a properly prepared specimen with a thickness that is small enough to be treated as a weak phase object. Additionally, optimum defocus conditions (Scherzer focus) must be applied. In recent years, the application of HRTEM to the study of crystal structures in various inorganic materials has become very common. However, the use of this technique for most polymeric materials has been limited by the rapid degradation of the material in the electron beam. 3.3.2 Evaluation and Reduction of Radiation Damage The radiation damage caused to polymers (see also Sect. .) during electron beam exposure depends on details of the interactions between the electrons and the molecules. Examples of these include chain scission, free radical formation, crosslinking and the destruction of crystal order. Usually, the sensitivity of polymers to electronbeam illumination is quantitatively expressed by the total end-point dose (TEPD) Je , which is the incident electron dose needed for the complete disappearance of all crystalline reflections in the electron diffraction pattern. For HRTEM investigation, it is important to restrict exposure at the sample to less than Je / to avoid imaging the damaged lattice []. A related quantity is the critical end-point dose Jc . Taking into account that the loss of intensity of a Bragg reflection as a function of dose can be described by an exponential decay function, Jc is defined as the characteristic dose at which the intensity of a Bragg reflection is reduced by a factor of e− . Either Je or Jc is a useful measure of the sensitivity of polymeric materials for HRTEM investigations. Kumar and Adams have found an exponential dependence of Jc on the melting temperature for a broad range of melting temperatures of polymers from  K up to  K []. In their very recent review of HRTEM of ordered polymers and organic molecular crystals, Martin et al. have shown an updated version of the graph showing the correlation between beam stability and thermal stability []. Furthermore, it is well known that Je and Jc are dependent on the accelerating voltage and on the temperature. According to the variation in the inelastic cross-section with the velocity v of the incident electrons, the TEPD can be assumed to be proportional to v  [], and this dependency has been roughly confirmed by experiments [–]. Therefore, a gain of about two in the TEPD can be obtained if the accelerating voltage is increased from  kV to  kV, but an increase to  kV results in a gain of only about three in the TEPD. Tsuji and Kohjiya have reported results revealing the temperature dependence of the TEPD

58

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

of various polymer crystals []. They measured the values of Je for an accelerating voltage of  kV at room temperature and at . K, and found for example Je = .   electrons per nm at room temperature and Je = .   electrons per nm at . K for PE. The enormous increase in the TEPD—a factor of —shows the effectiveness of so-called cryoprotection. According to the dependence of the TEPD on the accelerating voltage and on the temperature, there are two ways to extend the lifespan of a polymer crystal under electron beam irradiation in a TEM; namely, increase the high voltage and cool the specimen to a low temperature. HRTEM investigations of polymers in a modern cryogenic transmission electron microscope (e.g. JEM SFX from JEOL) can use both of these approaches by applying an accelerating voltage of  kV and cooling the specimen to the temperature of liquid helium [–]. Additionally, most modern TEMs serve so-called low-dose units (LDU) or minimum-dose systems (MDS) in order to reduce the illumination dose in the area of interest to a value needed for actual investigations. LDU and MDS enable the beam to be aligned and focussed at a location slightly away from the region of interest, where the beam is exclusively switched for image recording. 3.3.3 Application of HRTEM One of the more insidious problems encountered during HRTEM imaging is the effect of sample drift, as a drift of more than . nm during exposure can easily ruin a micrograph []. It must be taken into account that a mechanical stage drift occurs after specimen movement, and also that changes in imaging conditions can cause a drift due to hysteresis effects of the lenses. Furthermore, a sample drift is caused by the heating-up and charging-up of the polymer sample. The coating of the specimen with a thin carbon film and the utilisation of carbon- or metal-coated microgrids (perforated films) are recommended methods for suppressing the specimen drift []. Even if the experimental conditions are properly accommodated to the different requirements of HRTEM investigations of polymers and image recording is optimised, information stored in the micrographs will be buried due to a poor signal-tonoise ratio. Therefore, computer processing and computer simulations of the image contrast play an important role when interpreting experimental results in the field of HRTEM. This is illustrated in Fig. . for the lattice imaging of a poly(tetrafluoroethylene) (PTFE) film, as reported in []. The HRTEM micrograph (a) and the corresponding microdiffraction pattern (b) show the obtained experimental results. The information, which is buried in the noisy micrograph (a), is revealed in the Fourier-filtered image (c) and has been interpreted by the computer-simulated image (d). For further information, including the application of commercially available software packages, the reader is referred to the reviews [, , , , ]. All of the specific aspects of HRTEM of polymers are discussed in these reviews in more detail, the historical development in this field is reported, and a lot of examples and corresponding references are presented. Although HRTEM of polymers is not as common as that of

3.3 High-Resolution Transmission Electron Microscopy

59

Fig. 3.3. a Selected region of a HRTEM micrograph (100 nm defocus) and b the corresponding microdiffraction pattern; c is a Fourier filtered image reconstructed from the fast Fourier transform of (a) (shown in the inset) and (d) is a simulated image for similar operating conditions and an assumed sample thickness of 20 nm (the inset shows the 157 helical conformation used to obtain this image, not to scale). (Reprinted from [22] with the permission of Springer-Verlag)

inorganic materials, a lot of studies of polymer microstructures have provided several new and important insights. It has been established that, just like low-molecularweight solids, polymer crystals contain classical lattice defects such as dislocations and grain boundaries. Defect-mediated curvature and twisting in polymer crystals have been directly visualised by HRTEM []. The sizes, shapes, relative orientations and internal perfections of certain polymer crystals have now been determined [], and an analysis of displacement fields near dislocation cores in ordered polymers has also been carried out [].

60

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

3.4 Phase Contrast Transmission Electron Microscopy 3.4.1 Phase Contrast at Large Defocus Values As described above, the optimum conditions for phase contrast transfer, which enable to a high resolution to be obtained, occur at the Scherzer focus. However, low spatial frequencies do not significantly contribute to the image at this focus setting if they belong to the first part of the contrast transfer function, where the values of sin χ(u) are not sufficiently close to – (compare Fig. .b). Lamellar structures of semicrystalline polymers or periodic patterns of block copolymers are not imaged at the Scherzer focus if their long periods correspond with the low spatial frequency region outside the passband of spatial frequencies contributing to image contrast. As shown in Figure .c, the passband of useful values of u can be shifted towards lower spatial frequencies by a fairly large amount of underfocus. However, this is accompanied by a decrease in the bandpass width and also a shift of the first zero of sin χ(u) at uz , i.e. a loss in resolution. Nevertheless, this tuning of phase contrast at large defocus values can be used to image structures of interest. Applications of this phase contrast technique to image the morphology of untreated polymeric materials were first introduced by Petermann and Gleiter []. Later, detailed studies were carried out to reveal the use and misuse of the defocus electron microscopy of multiphase polymers [–], and recently the technique was successfully applied to directly image spherulites with a sheaf-like appearance in thin films of isotactic PS [, ]. Furthermore, a TEM equipped with a so-called Lorentz lens, which is a minilens in the lower pole piece of the objective lens, has been applied to image at large defocus values the structures of modifier particles in highimpact PS (HIPS) and lamellar structures of styrenic block copolymers []. A result of the latter is shown in Fig. .a. The unstained polystyrene-block-polyisopreneblock-polystyrene (SIS block copolymer) was recorded in Lorentz mode at a strong defocus of some tenths of a micron to visualise the lamellar structures by optimised phase contrast. 3.4.2 Phase Contrast by Means of Phase Plates In the case of optical observations, phase contrast optical microscopy has been widely applied by using a Zernike phase plate to provide phase contrast. Though there were different attempts to introduce this principle of phase contrast into TEM in the s, the Zernike phase contrast method has not yet been put into practice for transmission electron microscopy due to technical problems. Recently, however, such problems were overcome by a technique described in [, ]. The Zernike plate, which takes the form of a thin carbon film with a central hole about  μm in diameter, is positioned in the back-focal plane of the objective lens. The central unscattered electron beam passes through the hole, while the scattered electrons that pass through the film outside the hole are retarded in phase by Δφ = −π. This is realised by

3.4 Phase Contrast Transmission Electron Microscopy

61

Fig. 3.4. Lorentz micrograph (a) and electron phase image reconstructed from an electron hologram (b) of polystyrene-block-polyisoprene-block-polystyrene copolymer (SIS-S50) with a lamellar periodicity of 34–36 nm (a) and 29 nm (b). (Reprinted from [30] with the permission of Wiley)

using a Zernike plate of an appropriate thickness, which can be determined by the following relativistic formula for the additional phase delay Δφ []: Δφ = −π

hpl Vpl  + aV . λ V  + aV

(.)

Here, hpl and Vpl are the thickness and the inner potential of the Zernike plate, respectively, V is the accelerating voltage, λ is the electron wavelength and a = .  − V − is a constant. The phase delay Δφ = −π/ must be added to χ(u), and so, according to sin χ(u) −

π  = − cos χ(u), 

(.)

the sine function is replaced by the cosine function in the imaginary part of the contrast transfer function, which is responsible for the phase contrast in the image. The dotted-line plots in Fig. .a obviously show that there is a relatively broad passband beginning at u =  where cosχ(u) is close to . The resolution, again determined by the first zero of cosχ(u), is only somewhat worse than in the case of the Scherzer focus for high-resolution imaging. First applications of phase contrast imaging with the aid of a Zernike plate to polymeric materials are recently described in []. The experiments were carried out in a JEM-FFC TEM from JEOL equipped with a field-emission gun, a cryogenic specimen stage cooled with liquid helium, an energy filter and a CCD camera. The

62

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

objective lens was specially designed to accommodate the phase plate accurately on the back-focal plane. The reported results also include the application of another kind of phase contrast imaging. By replacing the Zernike plate by a semicircular phase plate, the so-called Hilbert differential contrast is produced []. Although this contrast is not directly related to the three-dimensional shape of the specimen, a quasitopographic image of the sample seems to appear.

3.5 Electron Holography 3.5.1 Introduction Phase contrast imaging by electron holography (see, e.g., [–]) provides a further, alternative means of recovery phase modulation in a TEM for weakly scattering systems such as unstained polymeric materials. Historically, in the first article of Gabor [], who later called the new method “holography” [], the actual purpose was to invent an electron-optical technique for strongly magnified images, the aberrations of which could be eliminated afterward by light optical processing. In electron holography, the phase shift of the electron wave due to the interaction with the specimen is determined by causing that wave to interfere with a reference wave of known phase and amplitude distribution. A lot of different schemes for producing electron holograms are known. According to the initial proposal of Gabor, electron shadow imaging techniques can be used to form the hologram. Here, the incident electron beam comes from a very small bright source which is close to the specimen, and the transmitted unscattered wave serves as the reference wave. The hologram is, in this case, a Fresnel diffraction image of the object. On the one hand, this kind of electron holography has been realised by a modern, simple experimental set-up, using a single-atom nanotip as the electron point source, a low working voltage (– V) and a piezomechanical nanodisplacement system for controlling both the position of the specimen and the tip–specimen distance. The results of applying this Fresnel projection microscopy to carbon and polymer fibres have been reported in []. On the other hand, the optics of the projection microscopy can also be realised in a modern STEM equipped with a field-emission gun, using a fixed and slightly defocused electron probe. However, this technique has not progressed beyond the demonstration of feasibility. 3.5.2 Image Plane Off-Axis Holography The most advanced holographic technique used today in a TEM is image plane offaxis holography. Here, the illuminating wave is split into the object wave and the reference wave. The reference wave ψref moves outside of the specimen through the vacuum, and so is not affected by the sample. The object wave propagates through the specimen, resulting in an object wavefunction ψob (x) at the exit surface of the specimen. Its subsequent modulation by the action of the objective lens via the transfer function is described in Sect. ... An electron biprism, i.e. a positively charged wire,

3.5 Electron Holography

63

is arranged behind the objective lens and superimposes the object and the reference waves in the image plane. The plane reference wave with an amplitude equal to  and a tilt corresponding to u = uc can be expressed in the following form: ψref = e  π i u c x .

(.)

Thus, the intensity distribution of the interference pattern (the so-called hologram) is given by the equation 

Ihol (x) = ψob (x)t(x) + ψref   =  + ψob (x)t(x) +  ψob (x)t(x) cos[πuc x + φob (x)]

(.)

where t(x) is the spread function described in Sect. .. and φob (x) is the phase of the object wavefunction. The first term “” and the second term correspond to the intensities of the reference and the image waves, respectively. Additionally, Eq. . contains the encoded image wave in the third term. This term describes a set of cosinoidal interference fringes at the given spatial carrier frequency uc with phase shifts φob (x) and amplitudes ψob (x)t(x) that represent, respectively, the phase and the amplitude of the image wave. Highly monochromatic and sufficiently coherent illumination by a field-emission gun is necessary to achieve a high contrast of the hologram fringes. Furthermore, taking into account that hologram fringes must have a spacing of less than one third of the achievable resolution limit, the electrical and mechanical stability of the instrument has to be very high. Usually, after being magnified by subsequent lenses of the TEM, the hologram is recorded by a CCD camera. Then the digitised hologram is fed into a computer, where the electron wave is reconstructed numerically by means of wave optical image processing techniques. This involves the application of a Fourier transform to the intensity distribution of Eq. ., resulting in F Ihol (x) = δ(u) + Ψi (u)T(u)Ψi (u)T  (u) + δ(u + uc )[Ψi (u)T(u)] + δ(u − uc )[Ψi (u)T  (u)]

(.)

where the symbols have the same meanings as in Sect. .. and conjugate complex quantities are marked with a star. In this expression, the first two terms (first row) form the central band, which agrees with the normal diffraction pattern of the specimen. The most interesting advantage over a conventional diffractogram is the additional information found in each of the sidebands formed by the third (second row) and fourth terms (third row) of Eq. .. They represent, properly isolated from the remainder, the complex Fourier spectra of the image wave and its conjugate complex, respectively. Thus, the spectrum is very convenient for further optical wave processing by means of a computer. If the third term is selected by using a suitable aperture and multiplied by T  (u), the resultant image is just ψob (x), corrected for aberrations. Thus, the real and imaginary parts of ψob (x), or its amplitude and phase components, can be derived separately. It should be noted that the phase component, and hence the phase contrast, is dominated by the real part of the spread function and accordingly by the cosχ(u) term [], as was described for Zernike phase contrast in the previous section.

64

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

3.5.3 Examples There are only a limited number of publications on electron holography of polymers, which all deal with styrenic materials. Using off-axis transmission electron holography, the mean free path for inelastic electron scattering in PS nanospheres has been determined [], and the particles could be imaged by quantitatively interpretable phase contrast []. Charged PS latex particles have been used to study electric charging at different image conditions with the aid of electron holography at low magnification []. Furthermore, results of electron holography investigations of HIPS and styrenic block copolymers have been reported recently [, , ]. An example from [] is given in Fig. .b. The micrograph shows the electron phase image reconstructed from an electron hologram of an unstained SIS block copolymer with lamellar morphology.

3.6 Low-Voltage Transmission Electron Microscopy 3.6.1 Introduction There is a decrease in resolution that is proportional to λ− (compare Eqs. . and .) if the accelerating voltage of the TEM is lowered. On the other hand, it is well known that the contrast in the TEM increases with decreasing electron energy and that this increase in contrast is significant if the accelerating voltage is on the order of kV. According to the results reported in [], an enhanced contrast at those voltages that is nearly  times higher than that for a voltage of  kV is observed, while the resolution decreases only threefold. Therefore, in the s and s, a few authors considered the construction of a low-voltage TEM (LVTEM) [–], but several technological problems blocked attempts to build a practically usable instrument. 3.6.2 A Dedicated Low-Voltage TEM and its Application However, taking advantage of the great progress made in the last few years in the construction of electron microscopes, a commercially available LVTEM (LVEM) has recently been developed by the Czech company Delong Instruments s.r.o. [, ]. The special goal of the company was to create an instrument which could conveniently provide enhanced image contrast for low atomic number specimens without a substantial loss of resolution. Based on a completely unconventional approach, the instrument comprises two basic parts: the first one is a miniaturised  kV TEM equipped with an ultrahigh vacuum chamber, a Schottky field-emission electron source, a simple electron-optical system and a fluorescent YAG screen, where the image is formed with a relatively low magnification (–). The second part is a standard light microscope (with several hundred times magnification) for the observation and recording of the image that appears on the fluorescent screen. The projection system, which consists of a YAG fluorescent screen and a light microscope,

3.7 High-Voltage Transmission Electron Microscopy

65

results in high image brightness, and so experiments can be carried out at low illumination dose. Furthermore, the LVEM is capable of operating in TEM mode and in STEM mode, and electron diffraction experiments can also be carried out. Recently, results have been reported which show that a broad range of unstained polymer samples, including polymer single crystals [], blends of both amorphous and semicrystalline polymers [–], block coploymers [] and electrospun nanofibres [], can be successfully imaged with the LVEM. However, not all mechanisms of image contrast formation are fully understood. Thus, an interpretation of contrast which can explain the differences found during the formation of contrast in imaged blends in LVTEM and LVSTEM is still the aim of further studies []. Furthermore, the relatively limited penetration capability of about  nm obtained for an accelerating voltage of  kV presents new challenges to sample preparation, which have been discussed in [–] for example. Additionally, it must be taken into account that the high image contrast obtained at low accelerating voltages results from a very strong interaction of the electron beam with the specimen, and this strong interaction, on the other hand, is also the main cause of the beam damage at those voltages.

3.7 High-Voltage Transmission Electron Microscopy 3.7.1 Introduction Reductions in the electron wavelength and in the scattering cross-sections are the most important effects of increasing the accelerating voltage, and so these effects have been the driving forces for the development of high-voltage transmission electron microscopes (HVTEMs), aimed at improving the resolution and the penetration power of the imaging electron beam. Additionally, the gain in lifetime (TEPD) obtained at increased accelerating voltages [–], as described in Sect. ., is a positive effect of HVTEM investigations of polymers. Due to the technical problems that arise with the increase in the accelerating voltage of a TEM, the development of highresolution TEMs for applications in materials science has mainly covered the intermediate voltage range of – kV so far. Therefore, the main goal of most of HVTEMs working at a voltage of  MV or even higher than this is the possibility of using specimen thicknesses that are some – times larger than possible with a conventional TEM. 3.7.2 Advantages and Applications of HVTEM Because of the low density of polymeric materials, specimens that are up to several micrometres in thickness can be investigated with sufficient resolution if an HVTEM with an accelerating voltage of about  MV is used [, –]. Relatively thick specimens are advantageous in several respects. On the one hand, it is much easier to prepare semi-thin sections from bulk material than ultrathin sections for investigations in a conventional TEM. Due to the significantly increased stability, the handling of thicker samples is also easier. On the other hand, the investigation of samples with increased thickness is of particular importance if structures in the micron

66

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

range (e.g. particles in HIPS) and their size distributions within the polymer material are to be studied. Micrographs produced by HVTEM investigations of semithin sections directly reveal the desired information, whereas conventional TEM investigations of ultra-thin sections give smaller size distributions, since smaller sections of the larger particles are imaged (the so-called “tomato salad” problem, see Chap. ) [, , ]. The large overall dimensions of an HVTEM (required for strong magnetic excitation of the lenses and for X-ray protection) provide an advantage during the installation of specimen-treatment devices for in situ experiments in which the specimen is (for example) heated, cooled or strained whilst under observation. In situ tensile tests which directly reveal deformation structures and mechanical microprocesses should preferably be carried out in an HVTEM, as the mechanical properties of polymer bulk materials are significantly better represented by those of semi-thin sections than by those of ultrathin sections. Furthermore, thicker tensile specimens can be prepared much more easily, as described in Chap. . The application of HVTEM to in situ tensile tests of polymers has been reviewed in [, –], and results from recent deformation experiments of different polymers with the aid of an HVTEM have been reported in [–]. Numerous examples are provided in this volume, particularly in Part III (see Chaps. , , ). It should be mentioned that the high velocities of the imaging electrons in an HVTEM usually cause the exposure time of photographic films to increase. Therefore, it is advantageous to use photographic material with a higher sensitivity for HVTEM investigations [].

3.8 Scanning Transmission Electron Microscopy 3.8.1 Introduction A scanning transmission electron microscope (STEM) combines some advantages of a TEM with those of a scanning electron microscope (SEM), which is the topic of discussion in Chap. . The electron beam is focussed to a spot that is as small as possible, and it is scanned across the specimen area to be investigated while the transmitted electrons are collected to form the signal. This combination has been realised in a very simple form for investigations of very thin specimens in conventional SEMs by making use of an attached detector for transmitted electrons or by detecting the signal of transmitted electrons with the aid of the secondary electron detector of the SEM. Features of the image contrast in the transmission mode of a SEM have been described in [] and the results of applying this special investigation technique to polymeric samples have been reported in, e.g., [, ]. However, common STEM investigations are carried out over a range of accelerating voltages that are typical of transmission electron microscopy. Most modern TEMs can easily be configured with scanning attachments, thus making them STEM as well as TEM instruments. STEM originally used electron microscopes specifically

3.8 Scanning Transmission Electron Microscopy

67

designed for scanning transmission imaging with the best resolution. Such a dedicated STEM with a field-emission gun was introduced by Crewe et al. [], and commercial versions of dedicated STEM instruments were available from the company Vacuum Generators (which is now no longer trading). 3.8.2 Similarities and Differences between STEM and TEM In many respects the STEM is similar to the TEM, while in others it is distinctly and even uniquely different. Thus, a dedicated STEM without a normal TEM capability does not need a set of post-specimen lenses to magnify the image or the diffraction pattern, as the transmitted electrons only need to be collected by an appropriate detector. On the other hand, STEM and TEM have an important aspect in common: applying the reciprocity principle to the ray paths of the microscopes, Cowley [] showed that there is a close relationship between STEM and TEM in terms of contrast formation. The different illuminations of the specimen in TEM mode and STEM mode (spot mode) are described in Sect. .. and illustrated in Fig. .. The incident electron beam in the specimen plane is almost a parallel beam of normal incidence in the TEM mode. The illuminating cones for each object point (illustrated in the more detailed ray diagram in Fig. .) have an illuminating aperture of the order of αi = .− mrad. In STEM mode, the electron beam is focussed to a spot in the specimen plane, resulting in a relatively high probe aperture (semi-angle of the cone of the focussed beam) of αp = − mrad. This value is the same as that of the objective aperture αol in the TEM mode, which is the semi-angle of the cone of electrons emanating from the same specimen point and entering the objective lens to be focussed at the corresponding image point. Therefore, there is an analogy between both microscope modes with respect to image formation if the electron detector acceptance angle αd of the STEM mode is chosen to be equal to the illumination angel αi of the TEM mode. Thus, a ray diagram that is completely identical to that of the TEM will result if the direction of the path of rays is reversed in the STEM. According to the reciprocity theorem, the contrast of a microscope is maintained if the electron source and detector are interchanged, i.e. if the path directions of the rays are reversed. This means that STEM and TEM show analogous contrast phenomena for αp = αol and αd = αi . Theoretical studies [] and practical demonstrations [] show that this analogy involves scattering, diffraction and phase contrast. However, it is only valid in the limit of the weakly scattering object approximation and at medium resolution. If inelastic scattering becomes appreciable, i.e. for increasing specimen thickness, reciprocity does not apply. Despite the analogy between the image processes in STEM and TEM due to the reciprocity principle, there are decisive differences between the image formation in both microscopes. In contrast to STEM, where the electrons pass through the lenses responsible for imaging before the electron–specimen interaction takes place, this process occurs after the interaction in TEM. Therefore, chromatic aberration due to the electron–specimen interaction, which can significantly limit the resolution in TEM mode imaging, has no influence on the image quality in STEM mode. Because of this advantage, STEM permits the investigation of thicker samples. Thus, the much

68

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

lower costs and far fewer installing difficulties associated with STEM can make it an advantageous alternative to an HVTEM. However, a loss in resolution due the topbottom effect caused by multiple scattering in thicker samples has to be taken into account. Further important advantages of the STEM mode are the production and positioning of small electron probes for X-ray analysis and electron energy-loss spectroscopy of small specimen areas. The results from an investigation performed by STEM and energy dispersive X-ray (EDX) analysis carried out at an accelerating voltage of  kV with a JEOL microscope of type JEM  equipped with a scanning unit and a Voyager II X-ray quantitative microanalysis system from Noran Instruments are shown in Fig. .. The STEM micrograph shows a segregated polystyrene/poly(styrene-co--bromostyrene) blend. The bromostyrene phase appears bright in the micrograph, as shown by the EDX line profiles from the K- and L-radiation of bromine in comparison with the grey level of the micrograph along the line, which is marked in the image by its dark contamination track. The additionally recorded line profile of the K-radiation of carbon does not show a correspondence with the grey-level profile, as it arose in both blend phases in the same way. The microscopic investigations of this polymer blend also directly revealed differences in the beam damage caused in STEM and TEM mode. While in STEM mode the segregated phases of the blend could be imaged in recorded micrographs and EDX investigations could also be carried out, when the same microscope was used in TEM mode the corresponding visualisation failed, as the original segregation disappeared in a very short time, even at low electron beam current densities. This experience corresponds to the results of comparative studies of beam damage in paraffin single crystals in STEM and TEM reported in []. The authors of these studies also found a significant increase in observation time for STEM investigations. 3.8.3 Application of Bright-Field and Dark-Field Modes A modern STEM instrument is usually equipped with an axial and an annular electron detector, resulting in two different contrast modes. The bright-field image is generated with the aid of the axial detector, which collects transmitted electrons and electrons scattered inside the small detector acceptance angle αd . If a sufficiently thin specimen is used, the reciprocity principle is valid and the image contrast resembles that of a conventional TEM, as described above. For thicker specimens, the detected signal corresponds to a large fraction of inelastically scattered electrons that are concentrated into a limited angular range. A dark-field image is preferably generated by means of a high-angle annular dark-field (HAADF) detector, which collects electrons scattered or diffracted into an angle greater than the axial detector acceptance angle αd . This high-angle signal mainly results from electrons that have experienced nuclear interactions. Therefore, the signal is incoherent and the contrast does not result from either diffraction or phase contrast. When high-angle elastic scattering (which occurs near individual

3.8 Scanning Transmission Electron Microscopy

69

Fig. 3.5. STEM and EDX investigations of a segregated PS/poly(styrene-co-4-bromostyrene) blend. STEM micrograph is shown with a dark contamination track after the EDX line scan and line scans along this track of the bromine K- and L-radiation, the carbon K-radiation and the grey-level of the STEM bright-field image

atoms) is utilised, the HAADF imaging mode has emerged as the primary imaging technique. STEM imaging in the HDAAF mode enjoys several advantages over conventional TEM imaging: () the spatial resolution is somewhat better; () it is sensitive to the atomic number of the imaged atoms; and () it provides a positive definite transfer of

70

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

the specimen’s spatial frequencies, allowing the direct interpretation of results with fewer ambiguities []. Using the HAADF image mode, a point-to-point resolution of better than . nm has been convincingly demonstrated with a  kV Cs -corrected STEM [, ]. In the field of polymers, this advantageous STEM imaging mode has been successfully applied for investigations of ionomers [–]. Typical ionomers are random copolymers with a minority of ionising monomeric units (typically acids) and a majority of nonionising monomeric units. The ionising groups can be partially or fully neutralized with ions. The resulting ionic groups microphase-separate from the nonionic monomeric units to create ionic aggregates with dimensions at the nanoscale. For example, a poly(ethylene-ran-methacrylic acid) melt neutralized with zinc (Zn-EMMA) contains randomly distributed spherical ionic aggregates with an average diameter of . nm, whereas a poly(styrene-ran-methacrylic acid) solution neutralized with caesium (Cs-SMMA) contains randomly distributed vesicular ionic aggregates with a diameter range from  to  nm and a shell thickness of about  nm []. Using a conventional TEM, nanoscale crystalline particles are usually imaged by positioning a particle in a Bragg condition and using diffraction contrast to generate an image. However, the nanoscale aggregates in ionomers are amorphous and their detection by mass-thickness contrast is an interpretive challenge because the phase contrast contributions of the surroundings have a similar appearance [, ]. However, the nanoscale aggregates in ionomers have been clearly imaged by STEM performed in both HDAAF mode [, , , ] and bright-field mode [, , ]. The additional use of tilt [] or double tilt [] enabled the series shapes, sizes and aspect ratios of the aggregates to be determined. Small aggregates (down to  nm) could be confidently detected and analysed by means of EDX []. Comparative studies of gold nanoparticles on PS support films revealed that the specimen thickness should not exceed  nm for optimum imaging of such small aggregates []. Furthermore, it has been reported that the application of image processing by deconvoluting raw STEM images of ionomers greatly decreases the noise level in the images, and can provide ways to improve investigations of the morphologies of ionomers [].

3.9 Analytical Transmission Electron Microscopy 3.9.1 Introduction Electron energy-loss spectroscopy (EELS) and X-ray microanalysis are the analytical techniques that can be used in TEM and STEM to investigate local elemental compositions. X-ray microanalysis in TEM and STEM mainly relies on EDX spectroscopy. X-ray quanta are emitted isotropically, but, owing to the restricted instrumental nature of a TEM, only a small solid angle (of the order of − sr) is collected by an EDX detector []. Furthermore, the fluorescence yield, i.e. the probability that a characteristic X-ray quantum is emitted rather than an Auger electron, is very low for light elements. For this reason, applications of this technique to polymer investigations like

3.9 Analytical Transmission Electron Microscopy

71

those presented in Fig. . have only rarely been reported. Therefore, X-ray microanalysis of polymers is not considered in this section, but it is described in Chap.  in conjunction with SEM investigations that generally yield better signal-to-noise ratios. EELS is the analysis of the energy distribution of electrons (with an initial energy E ) that have passed through the specimen and experienced an energy loss ΔE due to inelastic interactions with it. To discriminate the electrons according to their energy E = E − ΔE, magnetic fields are applied in electron energy-loss spectrometers which deflect incident electrons as a function of energy. The spectrometer is usually the final component of an analytical microscope. Unlike the emission of characteristic X-rays, the electron energy-loss spectrum does not show corresponding limitations for light elements. Another advantage of EELS is that electrons that have been inelastically scattered by ionisation processes are concentrated within small scattering angles, resulting in a correspondingly high collection efficiency of the spectrometer. There are two ways of detecting the electron energy-loss spectrum. A serial detector (serial EELS) uses a single detector and the spectrum is scanned across the slit in front of it by varying the strength of the magnetic prism, so that each energy is detected in turn. If a position-sensitive detector is used, the whole spectrum can be detected at once in a parallel spectrometer (parallel EELS, PEELS). Originally, the spectroscopic capabilities of EELS provided the basis for a highly effective materials analysis technique in TEM and STEM. Although EELS techniques have advanced notably in recent years, the development of imaging energy-filters and advances in corresponding instrumentation have largely made energy-filtered TEM (EFTEM) a powerful tool for the chemical analysis of materials at the nanometre scale. EFTEM offers, on the one hand, the mode of zero-loss filtering, where only the unscattered and elastically scattered electrons contribute to the image. On the other hand, a wide range of modes of electron spectroscopic imaging (ESI) and electron spectroscopic diffraction (ESD) result when only electrons that have energy losses corresponding to the selected energy-loss window are permitted to pass through the energy filter. This acquisition of maps that elucidate the spatial distribution of any feature in the electron energy-loss spectrum can be considered an extension of the classic EELS technique in two dimensions. There are currently two alternative EFTEM techniques in use. Dedicated EFTEMs with in-column filters (improved versions of the Ω-filter introduced by Rose and Plies []), for example the LEO  TEM shown in Fig. ., were developed by Zeiss and LEO and later by JEOL. Alternatively, post-column energy filters called Gatan imaging filters (GIFs) [] that can be attached as the final part of the electron-optical column of almost any TEM have been manufactured by Gatan. Both types of imaging energy filters can form images at a user-defined energy loss ΔE with electrons from a small energy range δE, which is determined by an adjustable slit aperture in the energy-dispersive plane of the spectrometer. If the slit aperture is omitted, all of the electrons contribute to the image (global mode), just as in a conventional TEM without an energy-loss filter. High spatial resolution electron energy-loss spectra (image

72

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

EELS) can be obtained from any position in the imaged specimen region by recording an extended series of images of the region (each image is successively shifted by ΔE) and subsequently reading the intensity as a function of ΔE from a particular position of interest. A scheme for an electron energy-loss spectrum is given in Fig. .. The plot of the inelastically scattered intensity as a function of the energy loss illustrates that a number of scattering processes contribute to the electron energy-loss spectrum, but it also shows a significant decrease in the mean intensity with increasing energy loss. Therefore, a corresponding intensity enhancement is necessary to visualise characteristic features of the spectrum. The zero loss is located at the origin of the spectrum and corresponds to unscattered and elastically scattered electrons (E = E , ΔE = ). Electrons which excite valence electrons, inter- or intraband transitions, or plasmons (collective oscillations of conduction-band and valence-band electrons) contribute to the low-loss region of the spectrum with  eV < ΔE <  eV. Beyond this low-loss region is a smoothly falling background with the ionisation edges of atoms whose absorption energies for inner-shell ionisation are reached in this region superimposed on it. The onset energies of these edges are element-specific, enabling one to perform elemental analysis. Each absorption edge has its own characteristic fine structure, and information on chemical bonding, molecular structure and dielectric constants may be obtained from a detailed study of this energy-loss near-edge structure (ELNES). The absorption edge maximum is usually followed by a tail of smoothly falling intensity due to the acceleration of bound electrons in the continuum. A quantitative interpretation of energy-loss spectra and element distribution images requires a knowledge of the local specimen thickness x. The spectrum itself yields the corresponding information according to following relation x = λi ln 

It . I

(.)

Here It is the total integrated spectral intensity and I is the intensity of the zero-loss peak. The total inelastic mean free path λi is a material constant that depends on the incident-electron energy and the collection acceptance angle. The following consideration is restricted to the features of electron energy-loss spectra and related investigation modes that are relevant for polymer investigations. Detailed descriptions of instrumentation and special EELS and EFTEM techniques are beyond the scope of this book, and readers interested in obtaining this information are referred to the broad assortment of recent books that cover this field, e.g. [–]. 3.9.2 EELS Investigations Using a TEM equipped with a post-column energy-loss spectrometer (Gatan PEELS  or GIF), electron energy-loss spectra were acquired with an energy resolution of about  eV by Varlot et al. [–] in order to investigate the low-loss regions and ELNES fine structure of poly(ethylene terephthalate) (PET) [, ], PS []

3.9 Analytical Transmission Electron Microscopy

73

Fig. 3.6. Scheme of an electron energy-loss spectrum

and poly(methyl methacrylate) (PMMA) [,]. The authors performed these EELS analyses of different polymers in order to evaluate the possibility of obtaining chemical information on polymers at the nanometre scale. It was possible to assign the ELNES fine structure in the acquired spectra to different chemical bonds in agreement with molecular orbit calculations and the results of corresponding X-ray absorption spectroscopy (XANES) experiments. Moreover, the experiments carried out with the specimen cooled to liquid nitrogen temperature showed highly visible changes in the spectra as soon as the electron dose exceeded  Cm− (about  e− nm− ) for PET,  Cm− for PS, and  Cm− for PMMA when a large probe size was used. These results confirm that PMMA is very sensitive to electron beam irradiation, although it was also found that a high dose rate in a nanometre-diameter electron beam is less destructive, and spectra from far less degraded PMMA could even be obtained at  Cm− in this way []. The O/C ratio in PMMA and peaks in the low-loss regions of PS and PET are quite sensitive to irradiation, and so their variations have been used to determine the critical electron dose for radiation damage. Sometimes the characteristic peaks of aromatic carbon bonds in the low-loss region are used as fingerprints for the corresponding polymer component in heterogeneous polymers in order to distinguish it from the other components. Using a VG HB STEM equipped with a Gatan model  PEELS detector, Hunt et al. [] visualised the different phases of a blend consisting of PS and low-density polyethylene (LDPE) be means of spectrum imaging and integration of the spectrum intensity at .  . eV. This spectral region between the large peaks of zero loss and plasmon loss covers the small peak which corresponds to the π  π  excitation of the phenyl ring of PS. LDPE with only saturated carbon bonds does not show an increased energy loss in this region. Imaging the blend by means of the raw data resulted in

74

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

micrographs with low contrast between the blend components. However, by using PS and LDPE reference spectra as basis functions for multiple least-squares fits to spectra with unknown compositions at each pixel in the spectral image, quantitative maps of the distributions of PS and LDPE in the blend have been obtained with good quality. Chou et al. [] studied the influence of RuO staining of PS on the peak of the π  π  excitation of the phenyl ring by means of a Philips CM  TEM/STEM with an attached Gatan model  PEELS detector. They found that the peak disappeared after staining, indicating that the stain had covalently reacted with the aromatic rings in the specimen. Furthermore, as in the case of OsO -stained dienes, RuO -stained aromatics appear more brittle than when unstained. Taken together, these observations suggest that RuO both opens the aromatic ring and serves as a crosslinker between adjacent aromatic structures, similar to the manner in which OsO is believed to crosslink unsaturated diene rubbers. The latter was confirmed by Ribbe et al. [] by a quantitative analysis of the staining reaction of OsO with a polyisoprene-block-polystyrene copolymer using EELS, as performed with a JEOL FX TEM equipped with a Gatan filter. Thin -nm sections were exposed to OsO vapour for various exposure times. Thereafter, electron energy-loss spectra of the films were acquired with an energy width of δE =  eV at the carbon K-edge at  eV and the oxygen K-edge at  eV. The ratio of the oxygen atoms N oxygen (i.e. absorbed OsO ) to the carbon atoms N carbon (i.e. polystyrene and polyisoprene) present in the selected sample area was determined in relation to the staining time using the following equation: oxygen

(α, δE) N oxygen σKcarbon (α, δE) IK . = N carbon σKoxygen (α, δE) IKcarbon (α, δE) oxygen

(.)

IK and IKcarbon are the background-corrected integrals within the energy window δE under the oxygen K-edge and the carbon K-edge, respectively, for a collection aperture angle α as determined by the angle of the limiting objective aperture. The oxygen partial inelastic scattering cross-sections σK and σKcarbon of the oxygen and carbon K-shells, respectively, can be estimated with the help of the hydrogenic model (see, e.g., []). It has been found that the oxygen content saturates at a value of around %, which is in good agreement with the predicted value for the double OsO reaction and clearly inconsistent with the % oxygen content predicted for the single OsO reaction, suggesting that crosslinking is the dominant mechanism during the staining procedure. Spatially resolved EELS investigations were used by Siangchaew and Libera [] to measure the interfacial width between the phase-separated components PS and poly(-vinylpyridene) (PVP) in a PS/PVP blend. The experiments were performed in a Philips CM FEG TEM/STEM equipped with a post-column Gatan model  PEELS detector. There was sufficient contrast to identify regions of interest in unstained specimens by taking images in the STEM mode by means of an annular darkfield detector. Position-resolved core-loss electron energy-loss spectra at the carbon K-edge ( eV) and the nitrogen K-edge ( eV) were acquired at high spatial

3.9 Analytical Transmission Electron Microscopy

75

resolution. The carbon/nitrogen ratio was determined in the same way as expressed for the oxygen/carbon ration in Eq. .. By taking into account the effect of the finite probe size with a corresponding deconvolution, the profile of the oxygen/carbon ratio across a PS/PVP interface was found to yield an interfacial width of . nm, which agrees well with neutron studies on PS/PVP lamellar block copolymer interfaces. 3.9.3 Electron Spectroscopic Imaging The selection of energy windows in different regions of the electron energy-loss spectrum results in different spectroscopic imaging modes. Values of the energy window δE of between  eV and  eV have typically been used for the ESI of polymeric materials. Zero-loss imaging does not yield element-specific information, but it is often applied to improve the image quality of bright-field images, as shown in [,] for example. Its aim is to remove all inelastically scattered electrons so that only the unscattered electrons of the primary beam and the elastically scattered electrons that pass through the objective aperture contribute to the image, and hence zero-loss imaging will enhance the contrast and improve the resolution by avoiding chromatic aberration. This is more serious for a thicker specimen and demonstrated in Fig. . by a stained -nm-thick acrylonitrile-butadiene-styrene (ABS) semi-thin section imaged in a LEO  TEM with a Ω-filter used in the global mode (a) and by zero-loss filtering (b). Elemental mapping by core-loss imaging at ionisation edges of the spectrum is the most important analytical application of energy-filtered images at large energy losses. As the elemental signal is superimposed on a high background in the spectrum, the background should be subtracted from the energy-filtered image to obtain

Fig. 3.7. TEM bright-field images of a stained 400-nm-thick ABS thin section produced by global mode (a) and zero-loss filtering (b). (From: Michler GH, Lebek W (2004) Ultramikrotomie in der Materialforschung. Carl Hanser Verlag, München. Reprinted with the permission of Carl Hanser Verlag)

76

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

a net elemental mapping. The standard procedure is to record, with a selected energy window δE, two images (A and B) below the edge and a third (C) beyond the edge of the ionisation energy of interest. The former two images are used to obtain a fourth image (D) by extrapolating to the same energy used in image C. Such an image extrapolation consists of calculating the grey-level value for each pixel, using the values measured from images A and B and the well-known exponential decay for the energy-loss spectrum background. Finally, the background image D is subtracted from the post-edge image C to obtain the digital difference image, revealing a map of the net elemental signal. Elemental mapping is a very useful mode for identifying individual phases in heterogeneous polymers. However, due to the strong decay in intensity with increasing energy loss, core-loss imaging at ionisation edges and for large energy losses requires a relatively high intensity of the incident electron beam. Thus, the application of this technique to most polymeric materials has been limited by the rapid degradation of these materials in the electron beam, and the effects of sample drift can also cause insidious problems. Therefore, it is often better to produce contrast between the phases in heterogeneous polymers via so-called structure-sensitive imaging []. This can be realised by taking a micrograph with an energy window δE in the spectral range between about  eV and  eV, i.e. close to but not reaching the carbon K-edge, where the scattering due to carbon atoms is at its minimum and a lot of other elements show increased intensity from the tails of their edges. This results in a dark-field-like structure-sensitive image with a resolution and a sensitivity that are superior to those of elemental mapping. Successful applications of elemental mapping and structure-sensitive imaging of polymeric materials for identifying phases in phase-separated block copolymers [–], blends [, ], multicomponent polymer networks [, ], latex particles and latex films [, –], segmented polyurethanes [] silicone oil modified ABS [], polymeric nanocomposites [, ], electrically conducting polymers [] and polymers for light-emitting diodes [,] have been reported. Results of an investigation of a poly (ferrocenyldimethylsilane-block-styrene) block copolymer (PFS/PS block copolymer) by means of ESI are shown in Fig. .. The experiments were carried out in an LEO  analytical TEM with an integrated Ω-filter at an accelerating voltage of  kV. Ultrathin sections of thickness  nm were cut at room temperature with the aid of a Leica UCT ultramicrotome from a solution-cast PFS–PS block copolymer film. Owing to the slow evaporation of the solvent and subsequent annealing at   C, phase-separation results in a morphology of well-ordered PSF cylinders in a PS matrix if the PS volume fraction of the diblock copolymer is ΦPS = %. The initial part of the energy-loss spectrum of the PSF–PS block copolymer is shown in image .(a), and for the purposes of comparison the same spectrum region of a -nm-thick PS film is presented in image (b). An energy window of δE =  eV was used to record the zero-loss image (c) and the electron spectroscopic image at ΔE =  eV (d) of the block copolymer. The zero-loss image is a bright-field image with mass-thickness contrast due

3.9 Analytical Transmission Electron Microscopy

77

Fig. 3.8. Initial parts of electron energy-loss spectra of 50-nm-thick ultrathin sections of a PFS–PS block copolymer (a) and a PS homopolymer (b); EFTEM investigation of the PFS–PS block copolymer: zero-loss image (c), electron spectroscopic image at 230 eV energy loss (d) and elemental mapping (f) by means of the iron L2,3 edge shown in the corresponding part of the energy-loss spectrum (e)

78

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

to the density differences between the PS and PSF phases of the block copolymer. While the repeat unit of PS consists of hydrogen and carbon atoms, i.e. of light elements only, that of PSF also contains one iron atom and one silicon atom. Thus, because of their increased elastic scattering, the PSF cylinders appear dark in image (c). Although, the iron M, edge and the silicon L, edge are not visible in the spectrum of the block copolymer, the spectroscopic image (d) recorded at an energy loss of  eV clearly reveals the block copolymer’s morphology via structuresensitive contrast. Furthermore, it has been found that, by recording a series of electron spectroscopic images at successively increasing energy losses, the bright-field contrast vanishes at an energy-loss of about  eV and a transition to the darkfield-like contrast takes place. The optimum dark-field contrast with inelastically scattered electrons was obtained for the spectral region between ΔE =  eV and ΔE =  eV, while the contrast disappeared beyond the carbon K-edge, as expected. Image (e) shows the area of the energy-loss spectrum of the block copolymer at around  eV, where the iron L, edge is detected. Additionally, the three windows A, B and C that were used to obtain the mapping of the net iron signal shown in image (f) have been marked. The mapping can be used to identify the morphological structures of the block copolymer, although the image quality of the structure-sensitive image (d) is superior to that of the mapping, as mentioned above. The continuous change in contrast obtained by switching the energy window between ΔE =  eV and ΔE =  eV can also be used for contrast tuning. This method is of special interest for imaging more thickly stained sections. It allows one to change the contrast of strongly stained structures relative to weakly or unstained areas of a section, so that all of the structures can be seen with appropriate contrast in one micrograph. A corresponding example of a copolymer of PE and polypropylene (PP) stained with RuO is described in []. While the boundaries between PE and PP cannot be clearly distinguished in either the unfiltered image or at ΔE =  eV, maximum contrast and good separation of the phases was obtained at ΔE =  eV.

3.10 Electron Tomography 3.10.1 Introduction Valuable information about the structure and the chemistry of materials in the micron and nanometre range can be obtained by applying different investigation modes of transmission electron microscopy. However, since TEM micrographs provide only two-dimensional (-D) projections of the three-dimensional (-D) object under investigation, the interpretation of the material’s structure can be ambiguous. This is why electron tomography was developed (see, e.g., [–]); it provides a means to reconstruct the -D structures of objects from a series of images taken at regular tilt intervals and usually recorded digitally on slow-scan CCD cameras. Whilst electron tomography has been used in the biological sciences for over  years (see, e.g., [, ]), it has only recently been applied to materials science [–],

3.10 Electron Tomography

79

and has proven useful for providing -D information about a variety of polymeric structures, e.g. microphase-separated structures in block copolymers [–] and blends [,,,], self-assembly of rodcoil copolymer nanostructures [] and material distributions in polymer nanocomposites [, ]. 3.10.2 Data Acquisition, Image Alignment and Reconstruction The acquisition of the tilt series is carried out by sequentially tilting the specimen about a single axis, usually from one extreme tilt to the other. As the goniometer of the TEM must provide a high enough tilt range of typically − or greater to minimise artifacts in the reconstruction process, TEMs with large-gap pole pieces are used for electron tomography investigations. Usually, a tilt series is acquired with equal angular increments of about  or  . On the other hand, an optimisation of the tilt increments has proposed by Saxon et al. [] in order to improve the overall resolution for a given electron dose. In principle, the acquisition of a tilt series can be performed manually. However, automated data acquisition and analysis procedures [, ] have been developed for dose-efficient data collection and improved image processing. The next generation of automation routines [–] based on computer-controlled precalibrated goniometers with highly reproducible movements have been adapted to TEMs and provide a user-friendly platform for all processing steps, including acquisition, alignment, reconstruction and analysis, resulting in high-throughput electron tomography. When tilting the sample, image shifts and defocus changes can occur. The best possible resolution in the final -D electron tomogram can only be achieved when the tilt series is aligned very accurately. Therefore, the individual images need to be shifted onto a common tilt axis, and alignment is needed to remove any residual shifts of the individual images with respect to each other, requiring spatial and rotational shift and the correction of scan or lens distortions. Generally, the alignment of the individual image is carried out by cross-correlating the images [] or by tracking fiducial markers within the images []. Combinations of both techniques have also been used. The former is usually a relatively uncomplicated and fast procedure, which however cannot correct for rotation or magnification changes between the individual images and is generally best-suited to nonshrinking specimens, the projected geometries of which do not change radically as a function of tilt []. An accurate marker-free alignment based on cross-correlation techniques with simultaneous geometry determination and reconstruction of tilt series has been reported recently []. Least-squares tracking of easily recognisable features in the images throughout the tilt series have been used as an alternative alignment technique. Usually, nanogold clusters deposited onto the specimen surface serve as markers, since most materials under investigation do not intrinsically contain fine globular particles which can be used as fiducial markers. Alignment data collected by means of fiducial marker tracking also contain information on changes in magnification and image rotation, and improved resolution of the reconstructed tomogram is obtained due to a more accurately aligned image stack.

80

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

After alignment of the data series, the -D reconstruction must be carried out. There are different reconstruction techniques in use. The most widely used technique is weighted back-projection (WBP), which is described in detail in [, ]. Weighting by a frequency filter back-projection involves projecting each -D image into a -D reconstruction space back along its original tilt angle. The superposition of the back-projected images results in the -D reconstruction of the specimen. The WKB technique is limited by noise and the loss of both high and low frequency information. Therefore, the application of iterative reconstruction techniques such as the algebraic reconstruction technique (ART) [] and the simultaneous iterative reconstruction technique (SIRT) [], which rely on iteratively optimising the -D tomograms, is preferable, in particular for the -D reconstruction of noisy and/or undersampled datasets [,]. Special computer software packages such as IMOD [, ] have been developed which can be used as tools for analysing and viewing the reconstructed -D image data. 3.10.3 Resolution of Reconstructed Data The resolution of the -D reconstruction of the specimen is significantly influenced by data collection. For the single-axis tilt geometry considered so far, the resolution parallel to the tilt axis is equal to the resolution of the recorded -D images. However, the resolution in the perpendicular direction is controlled by the number of projections acquired N. Assuming that the N projections cover the total tilt range of  , the resolution in those directions is inversely proportional to N. In practice, the tilt range is limited and cannot be extended to  . Due to this missing information, the resolution in the direction parallel to the electron-optical axis, i.e. parallel to the incident electron beam, is degraded by a corresponding elongation factor. This problem can be better understood by considering the relationship between a projection in real space and the Fourier space. The “central slice theorem” upon which computerised tomography relies states that the Fourier transformation of a projection at a given angle is a central section at the same angle through the Fourier transform of that object. As the unsampled volume in the Fourier space becomes wedge-shaped, the cause of the degraded tomogram due to the limited tilt range is called the “missing wedge”. A more detailed discussion of this problem has been given in [, ]. Here, only the conclusion is noted: to obtain maximum -D information, as many projections as possible should be acquired over a tilt range that is as wide as possible. However, problems arising from a limited tilt range for a single-axis tilt series can be solved by applying dual-axis electron tomography [,]. By using an additional second tilt series with its tilt axis perpendicular to that of the first series, the missing volume in the Fourier space can be significantly decreased by changing the “missing wedge” to a “missing pyramid”, resulting in a correspondingly improved reconstruction if both individual tomograms have been properly combined. Thus, the results reported in [] reveal that, in each individual reconstruction, the only cylindrical nanodomains of the block copolymer that were reproduced were the ones that were

References

81

properly oriented with respect to the tilt axis used, and also that complementary information is obtained by using two tilt series with orthogonal tilt axes. Combining the two sets of reconstructed data in Fourier space, on the other hand, results in an improved reconstruction where all of the cylindrical nanodomains are successfully captured, irrespectively of their orientations. Moreover, a significant enhancement of the image quality was observed in the -D reconstruction obtained from dual-axis tomography compared to that obtained from single-axis tomography. 3.10.4 Application of Electron Tomography The application of electron tomography relies on the assumption that the TEM operates as a projection tool that produces an image intensity distribution which fulfils the projection requirement. In principle, a monotonically varying function would be acceptable for a successful tomographic reconstruction []. Therefore, the mass-thickness contrasts of TEM bright-field images of biological and polymer samples fulfil the projection requirement very well. However, for the majority of specimens encountered in materials science (particularly crystalline materials), the TEM bright-field contrast depends on diffraction conditions, is caused by phase relations and does not show the required monotonic relationship to the amount of material through which the electron beam passes. To overcome this problem, special investigation techniques such as HAADF-STEM, HAADF-TEM and EFTEM are usually used when electron tomography needs to be applied in materials science (see, e.g., [, ]). Details of these new variants of electron tomography need not be described here, as polymer investigations are generally based on bright-field tomography. Nevertheless, the electron tomography of polymers takes advantage of advanced techniques; in particular, applications where EFTEM tomography is used to improve the quality of bright-field images by zero-loss imaging have been reported [, , ]. Although electron tomography has proven useful for revealing -D information about structures in polymeric materials [–], it should be noted that its application to polymers is mainly limited by the sensitivity of most of polymeric materials to beam damage, as the sample is extensively exposed to the electron beam during the acquisition of a tilt series, even when low-dose techniques are used.

References 1. Michler GH () Appl Spectrosc Rev : 2. Michler GH () Ultramicroscopy : 3. Dorset DL () Electron diffraction from crystalline polymers. In: Booth C, Price C (eds) Polymer characterization (Vol.  of Allen G, Bevington JC (eds) Comprehensive polymer science). Pergamon, Oxford, p  4. Dorset DL () Ultramicroscopy : 5. Dorset DL () Rep Prog Phys : 6. Voigt-Martin IG () Acta Polym : 7. Martin DC, Thomas EL () Polymer : 8. Kumar S, Adams WW () Polymer :

82

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

9. Martin DC, Chen J, Yang J, Drummy LF, Kübel C () J Polym Sci Polym Phys : 10. Reimer L () Transmission electron microscopy: physics of image formation and microanalysis, rd edn. Springer, Berlin 11. Kobayashi K, Ohara M () Voltage dependence of radiation damage to polymer specimens. In: Uyeda R (ed) Electron Microscopy  (Proc th Int Congr for Electron Microscopy). Maruzen Company Ltd., Tokyo, p  12. Richardson MJ, Thomas K () Aspects of HVEM of polymers. In: Electron Microscopy  (Proc th Eur Congr on Electron Microscopy). The Institute of Physics, Bristol, p  13. Salih SM, Cosslett VE () Some factors influencing radiation damage in organic substances. In: Sanders JV, Goodchild DJ (eds) Proceedings of the th International Congress on Electron Microscopy, vol . Australian Acad Sci, Canberra, Australia, p  14. Thomas EL, Ast DG () Polymer : 15. Martinez JP, Locatelli D, Balladore JL, Trinquier J () Ultramicroscopy : 16. Ohno T, Sengoku M, Arii T () Micron : 17. Tsuji M, Kohjiya S () Prog Polym Sci : 18. Tosaka M, Hamada N, Tsuji M, Kohjiya S, Ogawa T, Isoda S, Kobayashi T () Macromolecules : 19. Tosaka M, Tsuji M, Kohjiya S, Cartier L, Lotz B () Macromolecules : 20. Tosaka M, Kamijo T, Tsuji M, Kohjiya S, Ogawa T, Isoda S, Kobayashi T () Macromolecules : 21. Tsuji M () Electron microscopy. In: Booth C, Price C (eds) Polymer characterization (Vol.  of Allen G, Bevington JC (eds) Comprehensive polymer science). Pergamon, Oxford, p  22. Plummer CJG, Kausch H-H () Polym Bull : 23. Kübel C, González-Ronda L, Drummy LF, Martin DC () J Phys Org Chem : 24. Drummy LF, Voigt-Martin I, Martin DC () Macromolecules : 25. Petermann J, Gleiter H () Phil Mag : 26. Roche EJ, Thomas EL () Polymer : 27. Handlin DL Jr, Thomas EL () Macromolecules : 28. Handlin DL Jr, Thomas EL () J Mater Sci Lett : 29. Tsuji M, Fujita M, Shimizu T, Kohjiya S () Macromolecules : 30. Simon P, Rameshwar A, Lichte H, Michler GH, Langela M () J Appl Polym Sci : 31. Danev R, Nagayama K () Ultramicroscopy : 32. Nagayama K, Danev R () Microsc Anal : (EU) 33. Willasch D () Optik : 34. Tosaka M, Danev R, Nagayama K () Macromolecules : 35. Danev R, Okawara H, Usuda N, Kametani K, Nagayama K () J Biol Phys : 36. Tonomura A () Electron holography, nd rev edn. Springer, Berlin 37. Voelkl E, Allard LF, Joy DC () Introduction to electron holography. Plenum, New York 38. Lichte H, Lehmann M () Adv Imaging Electron Phys : 39. Gabor D () Nature : 40. Gabor D () Proc R Soc Lond Ser A : 41. Binh VT, Semet V, Garcia N () Ultramicroscopy : 42. Chou T-M, Libera M, Gauthier M () Polymer : 43. Chou T-M, Libera M () Ultramicroscopy : 44. Frost BG, Voelkl E () Mater Char : 45. Simon P, Huhle R, Lehmann M, Lichte H, Mönter D, Bieber T, Reschetilowski W, Adhikari R, Michler GH () Chem Mater : 46. Simon P, Lichte H, Drechsel J, Formanek P, Graff A, Wahl R, Mertig M, Adhikari R, Michler GH () Adv Mater : 47. Delong A, Hladil K, Kolařik V () Microsc Anal :(EU) 48. Wilska AP () Low voltage electron microscope. In: Houwink AL, Spit BJ (eds) Proceedings of the European Regional Conference on Electron Microscopy, vol . De Nederlandse Vereniging voor Electronenmicroscopie, Delft, p  49. Wilska AP () Low voltage electron microscope with energy filtering and amplitude contrast. In: Favard P (ed) Microscopie électronique , vol  (Proc th Int Congr on Electron Microscopy). Société Française de Microscopie Électronique, Grenoble, p  50. Heinemann K, Möllenstedt G (/) Optik : 51. Stolz H, Möllenstedt G () Optik :

References

83

52. Delong A () Low voltage TEM. In: Ríos A, Arias JM, Megías- Megías L, López-Galindo A (eds) Electron microscopy , vol  (Proc th Eur Congr on Electron Microscopy). Secretariado de Publicaciones de la Universidad de Granada, Granada, p  53. Delong A, Kolařik V, Martin DC () Low voltage transmission electron microscope LVEM-. In: Benavides HAC, Yacamán MJ (ed) Electron microscopy , vol (Proc th Int Congr on Electron Microscopy). Institute of Physics Publishing, Bristol, p  54. Lednický F, Coufalová E, Hromádková J, Delong A, Kolaøik () Polymer : 55. Lednický F, Hromádková J, Pientka Z () Polymer :  56. Lednický F, Pientka Z, Hromádková J () J Macromol Sci B : 57. Drummy LF, Yang J, Martin DC () Ultramicroscopy : 58. Michler GH () Progress in polymer investigations by high voltage electron microscopy. In: Sedlacek (ed) Morphology of polymers. Walter de Gruyter & Co, Berlin New York, p  59. Michler GH () Morphology of polymers. In: Bethge H, Heydenreich J (eds) Electron microscopy in solid state physics (Mater Sci Monogr ). Elsevier, Amsterdam, p  60. Michler GH () Kunststoff-Mikromechanik: Morphologie, Deformations- und Bruchmechanismen von Polymerwerkstoffen. Carl Hanser Verlag, München 61. Michler GH () Crazing in amorphous polymers: Formation of fibrillated crazes near the glass-transition temperature. In: Grellmann W, Seidler S (eds) Deformation and fracture behaviour of polymers. Springer, Berlin, p  62. Michler GH () Micromechanical mechanisms of toughness enhancement in nanostructured amorphous and semicrystalline polymers. In: Michler GH, Baltá-Calleja FJ (eds) Mechanical properties of polymers based on nanostructure and morphology. Taylor & Francis, Boca Raton, FL, p  63. Kim G-M, Michler GH, Rösch J, Mühlhaupt R () Acta Polym : 64. Laatsch J, Kim G-M, Michler GH, Arndt T, Süfke T () Polym Adv Technol : 65. Ivankova EM, Adhikari R, Michler GH, Weidisch R, Knoll K () J Polym Sci Polym Phys : 66. Ivankova EM, Krumova M, Michler GH, Koets PP () Colloid Polym Sci : 67. Seydewitz V, Krumova M, Michler GH, Park JY, Kim SC () Polymer : 68. Michler GH, Dietzsch () Cryst Res Technol : 69. Golla U, Schindler B, Reimer L () J Microsc : 70. Cudby PEF, Gilbey BA () Rubber Chem Technol : 71. Crewe AV, Wall J, Welter LM () J Appl Phys : 72. Cowley JM () Appl Phys Lett : 73. Zeitler E, Thomson MGR () Optik :, 74. Crewe AV, Wall J () Optik : 75. Krause SJ, Allard LF, Bigelow WC () STEM versus CTEM beam damage: paraffin single crystals. In: Sturgess JM (ed) Electron microscopy , vol  (Proc th Int Congr on Electron Microscopy). Microscopical Society of Canada, Toronto, p  76. Batson PE, Dellby N, Krivanek OL () Nature : 77. Krivanek OL, Nellist PD, Dellby N, Murfitt MF, Szilagyi Z () Ultramicroscopy : 78. Handlin DL, MacKnight WJ, Thomas EL () Macromolecules : 79. Laurer JH, Winey KI () Macromolecules : 80. Winey KI, Laurer JH, Kirkmeyer BP () Macromolecules : 81. Kirkmeyer BP, Weiss RA, Winey KI () J Polym Sci Polym Phys : 82. Kirkmeyer BP, Taubert A, Kim J-S, Winey KI () Macromolecules : 83. Taubert A, Winey KI () Macromolecules : 84. Kirkmeyer BP, Puetter RC, Yahil A, Winey KI () J Polym Sci Polym Phys : 85. Benetatos NM, Smith BW, Heiney PA, Winey KI () Macromolecules : 86. Rose H, Plies E () Optik : 87. Krivanek OL, Gubbens AJ, Dellby N, Meyer CE () Microsc Microanal Microstruct : 88. Reimer L (ed) () Energy-filtering transmission electron microscopy. In: Hawkes PW (ed) Advances in electronics and electron physics. Academic, Boston, MA 89. Reimer L (ed) () Energy-filtering transmission electron microscopy (Springer Ser Opt Sci ). Springer, Berlin 90. Egerton RF () Electron energy-loss spectroscopy in the electron microscope, nd edn. Plenum, New York 91. Shindo D, Oikawa T () Analytical electron microscopy for materials science. Springer, Tokyo 92. Ernst F, Rühle M (eds) () High-resolution imaging and spectrometry of materials. Springer, Berlin

84

3 Transmission Electron Microscopy: Conventional and Special Investigations of Polymers

93. Ahn CC (ed) () Transmission electron energy loss spectrometry in materials science and the EELS atlas, nd edn. Wiley-VCH, Weinheim 94. Varlot K, Martin JM, Quet C, Kihn Y () Ultramicroscopy : 95. Varlot K, Martin JM, Quet C, Kihn Y () Macromol Symp : 96. Varlot K, Martin JM, Quet C () J Microsc : 97. Varlot K, Martin JM, Gonbeau D, Quet C () Polymer : 98. Varlot K, Martin JM, Quet C () Micron : 99. Hunt JA, Disko MM, Behal SK, Leapman RD () Ultramicroscopy : 100. Chou TM, Prayoonthong P, Aitouchen A, Libera M () Polymer : 101. Ribbe AE, Bodycomb J, Hashimoto T () Macromolecules : 102. Siangchaew K, Libera M () Macromolecules : 103. Correa CA, Hage E Jr () Polymer : 104. Correa CA, Bonse BC, Chinaglia CR, Hage E Jr, Pessan LA () Polym Test : 105. Kunz M, Möller M, Cantow H-J () Makromol Chem Rapid Commun : 106. Kunz M, Möller M, Heinrich U-R, Cantow H-J () Makromol Chem Macromol Symp : 107. Cantow H-J, Kunz M, Klotz S, Möller M () Makromol Chem Macromol Symp : 108. Du Chesne A, Lieser G, Wegner G () Colloid Polym Sci : 109. Gerharz B, Du Chesne A, Lieser G, Fischer W, Cai WZ () J Mater Sci : 110. Du Chesne A () Macromol Chem Phys : 111. Tanaka Y, Hasegawa H, Hashimoto T, Ribbe A, Sugiyama K, Hirao A, Nakahama S () Polym J : 112. Ribbe AE, Hayashi M, Weber M, Hashimoto T () Macromolecules : 113. Ribbe AE, Okumura A, Matsushige K, Hashimoto T () Macromolecules : 114. Du Chesne A, Wenke K, Lieser G, Wenz G () Acta Polym : 115. Du Chesne A, Gerharz B, Lieser G () Polymer Int : 116. Amalvy JI, Asua JM, Leite CAP, Galembeck F () Polymer : 117. Galembeck F, Leite CAP, da Silva MCVM, Keslarek AJ, Costa CARC, Teixeira-Neto E, Rippel MM, Braga M () Macromol Symp : 118. Rippel MM, Leite CAP, Galembeck F () Anal Chem : 119. Rippel MM, Leite CAP, Lee L-T, Galembeck F () Colloid Polym Sci : 120. Eisenbach CD, Ribbe A, Günter C () Macromol Rapid Commun : 121. Heckmann W, McKee GE, Ramsteiner F () Macromol Symp : 122. Amalvy JI, Percy MJ, Armes SP, Leite CAP, Galembeck F () Langmuir : 123. Lieser G, Schmid SC, Wegner G () J Microsc : 124. Lieser G, Oda M, Miteva T, Meisel A, Nothofer H-G, Scherf U () Macromolecules : 125. Loos J, Yang X, Koetse MM, Sweelssen J, Schoo HFM, Veenstra SC, Grogger W, Kothleitner G, Hofer F () J Appl Polym Sci : 126. Frank J (ed) () Electron tomography: three-dimensional imaging with the transmission electron microscope. Plenum, New York 127. Frank J () Three-dimensional electron microscopy of macromolecular assemblies. Academic, New York 128. Weyland M, Midgley PA () Materials Today : 129. Koster AJ, Ziese U, Verkleij AJ, Janssen AH, de Jong KP () J Phys Chem B : 130. Midley PA, Weyland M () Ultramicroscopy : 131. Spontak RJ, Williams MC, Agard DA () Polymer : 132. Spontak RJ, Fung JC, Braunfeld MB, Sedat JW, Argard DA, Kane L, Smith SD, Satkowski MM, Ashraf A, Hajduk DA, Gruner SM () Macromolecules : 133. Laurer JH, Hajduk DA, Fung JC, Sedat JW, Smith SD, Gruner SM, Agard DA, Spontak RJ () Macromolecules :  134. Jinnai H, Nishikawa Y, Spontak RJ, Smith RJ, Agard SD, Hashimoto T () Phys Rev Lett : 135. Jinnai H, Kajihara T, Watashiba H, Nishikawa Y, Spontak RJ, () Phys Rev E :- 136. Yamauchi K, Takahashi K, Hasgawa H, Iatrou H, Hadjichristidis N, Kaneko T, Nishikawa Y, Jinnai H, Matsui T, Nishioka H, Shimizu M, Furukawa H () Macromolecules : 137. Jinnai H, Nishikawa Y, Ikehara T, Nishi T () Adv Polym Sci : 138. Takano A, Wada S, Sato S, Araki T, Hirahara K, Kazama T, Kawahara S, Isono Y, Ohno A, Tanaka N, Matsushita Y () Macromolecules : 139. Kaneko T, Nishioka H, Nishi T, Jinnai H () J Electron Microsc : 140. Sugimori H, Nishi T, Jinnai H () Macromolecules : 141. Sengupta P, Noordermeer JWM () Macromol Rapid Commun :

References

85

142. Jinnai H, Hasegawa H, Nishikawa Y, Sevink GJA, Braunfeld MB, Agard DA, Spontak RJ () Macromol Rapid Commun : 143. Radzilowski LH, Carragher BO, Stupp SI () Macromolecules : 144. Wilder EA, Braunfeld MB, Jinnai H, Hall CK, Agard DA, Spontak RJ () J Phys Chem B : 145. Ikeda Y, Katoh A, Shimanuki J, Kohjiya S () Macromol Rapid Commun : 146. Saxton WO, Baumeister W, Hahn M () Ultramicroscopy : 147. Dierksen K, Typke D, Hegerl R, Koster AJ, Baumeister W () Ultramicroscopy : 148. Fung JC, Liu W, de Ruijter WJ, Chen H, Abbey CK, Sedat JW, Agard DR () J Struct Biol : 149. Ziese U, Janssen AH, Murk J-L, Geerts WJC, Van der Krift T, Verkleij AJ, Koster AJ () J Microsc : 150. Schoenmakers R, Perquin R, Fliervoet T, Voorhout W, Schirmacher H () Microsc Anal :(EU) 151. Nickell S, Förster F, Linaroudis A, Del Net W, Beck F, Hegerl R, Baumeister W, Plitzko JM () J Struct Biol : 152. Winkler H, Taylor KA () Ultramicroscopy : 153. Frank J, McEwen BF () Alignment by cross-correlation. In: Frank J (ed) Electron tomography: three-dimensional imaging with the transmission electron microscope. Plenum, New York 154. Lawrence MC () Least-squares method of alignment using markers. In: Frank J (ed) Electron tomography: three-dimensional imaging with the transmission electron microscope. Plenum, New York 155. Rademacher M () J Electron Microsc Tech : 156. Rademacher M () Weighted back-projection methods. In: Frank J (ed) Electron tomography: three-dimensional imaging with the transmission electron microscope. Plenum, New York 157. Gordon R, Bender R, Herman GT () J Theor Biol : 158. Gilbert P, () J Theor Biol : 159. Kremer JR, Mastronarde DN, McIntosh JR () J Struct Biol : 160. Boulder Lab For -D Electron Microscopy of Cells () The IMOD home page. http://biod.colorado.edu/imod/index.html 161. Mastronarde DN () J Struct Biol : 162. Hawkes PW () The electron microscope as a structure projector. In: Frank J (ed) Electron tomography: three-dimensional imaging with the transmission electron microscope. Plenum, New York

4 Scanning Electron Microscopy (SEM)

Today, scanning electron microscopy (SEM) is a versatile technique used in many industrial labs, as well as for research and development. Due to its high lateral resolution, its great depth of focus and its facility for X-ray microanalysis, SEM is often used in materials science – including polymer science – to elucidate the microscopic structure or to differentiate several phases from each other. After a brief historic overview, this chapter explains the assembly and the mode of operation of SEM, which deviates from standard microscopes. This includes descriptions of the fundamentals of electron optics, the electron optical column, and the physical basics of electron–specimen interactions, which aid the understanding of contrast formation and charging effects. Because it is important to know the factors that influence X-ray microanalysis, a separate section about the origins of X-ray spectra and their interpretation has also been added. A discussion of environmental scanning electron microscopy (ESEM™) – a special development of SEM that is particularly useful when nonconducting or “wet” samples are to be examined – completes the chapter.

4.1 A Brief History of SEM Parallel to the development of the transmission electron microscope (Sect. .), Max Knoll had the idea of developing an electron microscope for investigating compact or bulk samples, and he demonstrated the basic principle for a scanning electron imaging device using two Braun-type cathode ray tubes in  []. In both of these tubes, the electron beam is scanned by a pair of magnetic coils. In one of them, a plate with different surface layers was irradiated with a scanning beam, and the emitted secondary electrons that reached the anode were collected. The signal obtained was used to control the local brightness of the beam in the second tube. The surface layers of the irradiated plate could therefore be visualised by the second tube. This experiment was a proof of the material-specific dependence of secondary electron emission, but it was still not adequate for microscopic imaging. Manfred von Ardenne developed an electron probe microscope in  that used electron-optical lenses to focus the beam. This idea originated from the fact that resolution-limiting chromatic failures do not play a role in such a microscope [–]. He presented this instrument, in which electrostatic scanning plates were applied to a TEM, only a few years after Knoll’s and Ruska’s first TEM. This type of microscope is

88

4 Scanning Electron Microscopy (SEM)

now known as a scanning transmission electron microscope (STEM). However, Ardenne’s prototype fulfilled the minimum criteria for a scanning electron microscope (SEM) from today’s point of view. The first scanning electron microscope of the usual type was described and developed by Zworykin et al. [] in  using three electrostatic lenses and electromagnetic coils between the second and third lenses. Unfortunately there was no further development in the field of SEM in Germany, and Ardenne’s instruments were lost in  in an air raid during World War II. In the latter half of the s, Charles Oatley continued research into SEM at Cambridge University, but it was still quite some time until the first commercial SEM was made available. One of his graduate students, McMullan, enhanced the instrument of Zworykin in order to achieve a lateral resolution of about  nm []. Smith [] replaced the electrostatic lenses by electromagnetic ones and introduced a double deflection system to correct astigmatism. At the beginning of the s, Everhardt and Thornley [] improved the secondary electron detector by adding a light pipe between the scintillator and multiplier. All these improvements provided a great impetus for the development of the first commercial scanning electron microscope, which was named Stereoscan Mark  and sold by Cambridge Instruments in . Parallel to the Cambridge group, Coslett’s group at the Cavendish Laboratory combined the X-ray analytical capability of Castaing’s “microsonde electronique” (electron microprobe) [] with an imaging facility and developed the scanning electron probe X-ray microanalyser []. From the middle of the s until the present several improvements in electron generation and electron optics have been made that have enhanced the resolution power. At the end of the s, computerised SEM entered the scene, which led to the introduction of the digital scanning generator for digital image recording and processing. At that time a new type of SEM with a vacuum of some tens of mbar in the specimen chamber was developed. One reason for this development was the idea of directly preventing the charging of insulating samples; another reason was to make it possible to investigate humid samples. Because of the high-voltage flashovers of the Everhardt-Thornley detectors normally used, most instruments only allow imaging with backscattered electrons in the low vacuum mode. Danilatos provided a completely new principle, the “gaseous secondary electron detector” (GSED) []. Using this type of detector and subsequent developments from Danilatos [–], the newly formed ElectroScan Corporation produced the first “environmental scanning electron microscope” (ESEM™) in , which was based on electron optics from Philips Eindhoven that had the ability to image with secondary electrons. After the acquisition of ElectroScan in  by Philips/FEI, a new generation of ESEMs came onto the market. The special features of this instrument are explained in Sect. ..

4.2 Fundamentals of Electron Optics and Signal Generation 4.2.1 Principle of SEM The basic principle of the SEM [–] is based on Knoll’s experiment [] and von Ardenne’s idea for a scanning probe transmission microscope [].

4.2 Fundamentals of Electron Optics and Signal Generation

89

Fig. 4.1. Scheme showing the principle of SEM

In a similar process to the scanning of the electron beam in a cathode ray tube (CRT), the focussed electron beam scans line by line over the surface of the specimen in the evacuated microscope column and forms signals based on the interactions between the beam and the sample, which are electronically detected and amplified by suitable equipment. Originally, the response signal was displayed as a brightness modulation on a CRT where the electron beam is driven simultaneously to the beam in the column, as illustrated in Fig. .. Nowadays, digital computer techniques have replaced the traditional CRTs. As the area of the displayed image remains unchanged, the magnification of the image is determined by the dimension of the scanned sample area (see Sect. ..). Generally, the resolution of the SEM image is determined both by the diameter of the electron probe focussed on the sample surface (see Sect. ..) and the interaction of the primary electrons (PE) with the sample (see Sect. ..). 4.2.2 The Lateral Resolution Power of SEM Using the electron-optical system within the SEM column, the electron beam created by the electron gun (see Sect. ..) is reduced to a sufficient diameter to form a very fine focussed electron probe. The PE beam diameter in the electron gun depends on the type of cathode (respective gun type) and is formed by the first crossover of the electron trajectories (see Fig. .d). The first crossover generated from the tungsten cathode must be scaled down to approximately a factor of  to attain a reasonable resolution. Two or three (although sometimes more) electromagnetic lenses are needed to demagnify the beam diameter (see Fig. .). Furthermore, the first lens or the first and second lenses (called condenser lenses) are used to vary the PE beam current.

90

4 Scanning Electron Microscopy (SEM) Fig. 4.2. Scheme of the electronoptical relationship in SEM

The minimal achievable diameter d of the beam can be estimated with the help of the lens law (see Eqs. . and .). Another quantity used to characterise the electron beam performance is the gun brightness R, defined by R=

jc jo = = const.  πα c πα 

(.)

where j c is the beam current density at crossover, α c is the aperture angle at crossover, j  is the beam current density at the specimen surface and α  is the aperture angle at the specimen. The aperture angle α  is determined by the ratio of the diameter of the last diaphragm r to the working distance L, which is the distance between the pole piece of the objective lens and the specimen: α =

r . L

(.)

A homogeneous beam density within the beam diameter d at the sample surface corresponds to the following beam current I : I =

π  d j .  

(.)

4.2 Fundamentals of Electron Optics and Signal Generation

91

Combining Eqs. . and . results in I =

π   Rd α  

(.)

and d = 

I   C  = .  π R α α

(.)

This equation is also valid when the homogeneous illumination is replaced with a Gaussian-distributed one. In this (more realistic) case, d corresponds to the halfwidth of the full maximum (HWFM) of the distribution. It can be seen from Eq. . that the diameter of beam is directly proportional to the beam current and indirectly proportional to gun brightness and aperture angle. In reality, the effective diameter of the beam db at the specimen is somewhat larger due to aberrations (see Sect. ..). An acceptable signal-to-noise ratio in the secondary electron signal (see Sect. ..) usually requires a minimum beam current I of between − A and − A. Figure . shows the dependence of db on the aperture angle and the beam current for a tungsten filament gun. The figure also shows that there is an optimum value for the aperture angle as a function of the beam current. The gun brightness R of the Schottky emitter or field-emission guns (FEG) is higher than that of the tungsten or LaB cathodes (see Table .). According to Eq. ., a much lower beam diameter is achieved at the same beam current with the FEG rather than thermionic guns. Thus, if an FEG is used, a resolution power of  nm or even less can be achieved.

Fig. 4.3. Dependence of the effective beam diameter db (in the nm range) on the aperture angle α0 (in the mrad range) and the beam current I0 for a tungsten filament gun (schematic, after [15])

92

4 Scanning Electron Microscopy (SEM)

4.2.3 Comparison of Various Cathode Types The types of cathodes (usually termed “guns”) frequently used in an electron microscope have been already described (in Sect. .. and Table .). To select the most appropriate gun type, the operator should take into account first the minimum resolution power (Sect. ..) required for the investigations and secondly all of the other parameters of cathodes. Table . shows the main differences between tungsten and LaB cathodes and the FEG. While the beam diameter increases with increasing probe current for the first two types, in the case of FEG there is a probe current interval where the beam diameter stays constant with increasing probe current. The most important feature of the gun for the user of an SEM is the available probe current of the emitters. For certain SEM techniques, like WDX analysis (see Sect. ..) or cathodoluminescence, thermionic emitters have been preferred in the past. Nowadays thermal FEG is a good compromise. This emitter type offers on the one hand a good resolution in the SE mode and on the other hand sufficient probe current for the application techniques mentioned before. 4.2.4 Depth of Focus The depth of focus is one of the outstanding features of SEM. While the depth of focus of a light microscope for a magnification of about  is in the μm region, the depth of focus of the SEM at same magnification covers mm. The depth of focus is a function of both the convergence angle α  of the electron beam and the magnification M (Fig. .). Noting that tan α  α  at α  ll, the focus depth S is expressed by: S=

r . mm . = α α ċ M

(.)

Here, r is the beam diameter which produces a corresponding spot on the display which can just about be resolved by the human eye (ca. . mm). Equation . reveals that the depth of focus decreases with increasing magnification and aperture angle α  . The facet eye of a fly (see Fig. .) demonstrates high depth focus in SEM imaging. 4.2.5 Interaction of Primary Electrons with Sample Signal generation in SEM is a result of the interaction between the incident electron beam and a thin surface layer of the sample, which depends on the beam energy. The primary electrons are charged particles and so they interact strongly with the electrically charged particles of the atoms in the sample, i.e. both with negatively charged electron clouds and positively charged nuclei. The interaction is said to be inelastic if some of the energy of the primary electron is lost during the interaction. If no energy is lost the interaction is said to be elastic. The fundamentals and applications of these scattering processes are also described in Sects. .. and ..

4.2 Fundamentals of Electron Optics and Signal Generation

93

Fig. 4.4. Scheme of the derivation of depth of focus in SEM (after [15])

Fig. 4.5. SE micrograph of the facet eye of a fly showing a high depth of focus

94

4 Scanning Electron Microscopy (SEM) Fig. 4.6. Rutherford scattering of a primary electron in the Coulomb field of an atomic nucleus

PE/Nucleus Interaction When electrons enter the electric field of an atomic nucleus, which can be partly screened by electron clouds, their paths are deflected. Generally, this so-called Rutherford scattering results in parabolic PE trajectories. The larger the atomic number of the atoms in the interaction region and the smaller the distance between the nucleus and the PE, the stronger the deflections of the latter and hence the more curved PE paths. While contrast-forming interaction processes in a TEM are restricted to weak scattering events with small scattering angles (forward scattering), electron scattering in bulk samples in SEM also includes a high proportion of strong scattering events where the directions of the PE trajectories are significantly changed, with some even sent back along their original paths (back scattering), as illustrated in Fig. .. While a negligible amount of energy is lost due to conversion into X-ray continuum radiation (“bremsstrahlung”; see Sect. ..), the interactions are assumed to be elastic. Electrons that undergo one or more Rutherford processes and leave the sample surface without notable energy loss are called backscattered electrons (BSE). PE/Electron Cloud Interaction PEs striking the sample surface may also interact with the electrons of atoms within the surface region. Owing to their equivalent masses and identical charges, the interaction is inelastic. If the energy of the incident electron is high enough, the valence electrons of the surface atoms can easily be released from the atoms (Fig. .). These electrons are called secondary electrons (SE). Their kinetic energies are very low (only a few electron volts), just enough to surmount the work function of the sample. However, only SE originating from a very thin surface layer a few nm in thickness can contribute to the detectable signal, as all of the SEs generated in deeper regions of bulk samples will recombine. Generally, ionised or excited atoms will rapidly con-

4.3 The Instrumentation of SEM

95

Fig. 4.7. Inelastic interaction between primary electrons and electrons in atomic shells

vert back to their initial states. These processes are accompanied by the emission of characteristic X-rays (see Sect. ..) or Auger electrons, which is the basis for analytical investigations by means of SEM.

4.3 The Instrumentation of SEM 4.3.1 The Column Generation of the Focussed Electron Beam As already described in Sect. .., the SEM consists of a column (which is a unit containing the lens system that forms the finely focussed electron beam), a specimen chamber, detectors, as well as imaging and recording units (see Fig. .). The typical acceleration voltage range of SEM lies between some hundreds volts and  or  kV. PEs originating from the virtual source at the first crossover (in the electron gun) as a divergent beam pass through the anode aperture and enter the lens system of the SEM. Usually, a spray aperture is placed in the entrance plane of the lens system to block electrons moving along paths very far from the optical axis. The lens system acts on the electron beam in two ways. On the one hand, it transfers the PEs from the gun crossover to the plane of the specimen surface, and in doing this it reduces the beam diameter considerably to form a very fine probe. On the other hand, the beam current must be controlled by the lens system. In principle, a system of just two lenses combined with an aperture diaphragm could be used to do this, as

96

4 Scanning Electron Microscopy (SEM)

illustrated in Fig. .. The first lens, called a condenser lens, that has a variable and relatively weak magnetic field (and a correspondingly large focal length) is mainly used to control the beam current, while the second one, called the objective lens, is a short-focal lens (with a strong magnetic field) which is primarily responsible for the demagnification of the beam diameter. However, the condenser lens is usually replaced by a system of condenser lenses that allow the beam current and beam diameter to be varied independently. The aperture angle α  limiting diaphragm is usually placed within the pole piece of the objective lens (see Figs. . and .). This aperture can usually be adjusted from outside and must be corrected after changing the high voltage and beam current. One should note that the middle points of the images remain unaltered during the focussing operation. Scanning Unit and Stigmator Two pairs of scanning coils to deflect the electron beam in the x- and y-directions are located within the objective lens (see Fig. .). These coils are driven by two different sweep signals, where, as in a CRT, the slew rate of the pair of scanning coils for x-deflection is a multiple value of that for the other pair. The increase in the sweep rate over time influences the change in the magnetic field. Due to the change in the magnetic field in the coils, the electron beam will be deflected away from the optical axis and so the beam scans across the sample surface. In the process both of the sweep voltages range from a maximum negative value to the same positive value, so that the middle of the scanned area on the specimen surface occurs along the optical axis. The higher the maximum sweep voltage, the stronger the magnetic field in the coils and the larger the maximum elongation at the sample. The size of the scanned area on sample surface can be changed by varying this maximum voltage. This procedure controls the magnification of the SEM (Sect. ..). As mentioned in Sect. .., astigmatism occurs in all electron lenses due to instrumental imperfections introduced during the manufacturing process. In SEM, this astigmatism also leads to a deformation of the ideal spherical cross-section of the electron beam. Due to the deflection of the beam during the scanning process, the beam spot presents an increased elliptical cross-section, in particular at the fringes of the frame. This elliptical distortion leads to loss of resolution and to reduced image

Fig. 4.8. Scheme of an objective lens with scanning coils and stigmator

4.3 The Instrumentation of SEM

97

Fig. 4.9. SE image demonstrating the appearance of an elliptical focus due to astigmatism (left) and compensated image (right) (test sample Au on C)

sharpness. However, the so-called stigmator (see Fig. .) can be used to compensate this distortion (Fig. .). 4.3.2 Specimen Chamber and Goniometer A specimen chamber with a goniometer unit, which enables sample movement to be defined, is attached to the microscopic column below the pole piece of the objective lens. Usually, the goniometer not only enables rotations and translations of the specimen in all (x, y and z) directions, but it also enables the specimen to be tilted. The maximum tilt angle is usually  towards the Everhardt-Thornley detector (see below) and − in the opposite direction. Modern goniometers allow specimen movements of more then  cm in the x- and y-directions and a few cm along the z-direction. For a constant deflection of the electron beam, the scanned surface area depends on the distance between the specimen surface and the scanning coils (theorem on intersecting lines), i.e. z-translation is very important when choosing the optimal working distance (WD). In general, WD is defined as the distance between the pole piece of the objective lens and the specimen surface. It should be noted that a higher resolution can be achieved at lower WD. Usually a corresponding decrease in the detected SE signal due to shadowing effects limits the reduction of WD in practice. The use of an optimal WD is of special importance for analytical investigations performed by X-ray microanalysis (Sect. .). Often the sample is transferred to the specimen support of the goniometer via an airlock; otherwise the chamber and the lower part of the column have to be vented. The specimen support of the goniometer is electrically insulated from the other parts of the microscope. Thus, absorbed electrons can be detected using a special amplifier or a warning signal can be generated using a suitable electronic unit when the specimen collides with the microscope. It is also possible to incorporate special devices (heating or cooling units, tensile or bending modules, etc.) into the specimen chamber to carry out in situ tests of the sample (see Chap. ).

98

4 Scanning Electron Microscopy (SEM)

4.3.3 Detectors Secondary Electron Detector An SEM is normally equipped with an Everhardt-Thornley detector [] for imaging the sample surface via collected secondary electrons. This detector is a combination of a scintillator and a photomultiplier. Due to its low noise, a photomultiplier is favoured over other kinds of amplifiers used to amplify small currents in the range of nA to pA, which are typical of the emitted SE currents in SEM. The Everhardt-Thornley detector is based on the following principle. The secondary electrons that leave the sample possess energies of up to only  eV. These low-energy electrons can be collected with high efficiency by a grid electrode that is positively biased with a voltage of about  V. The collected SEs are subsequently accelerated toward the scintillator, which is covered with a thin aluminium layer and placed at a voltage of about + kV. The accelerated SEs striking the scintillator possess sufficient energy to emit photons by converting their kinetic energies. The generated photons pass through a light pipe into the photomultiplier, where they cause the emission of photoelectrons. The latter are highly amplified by electron multiplication at the dynodes in the multiplier, and so finally an electronic signal that is proportional to the number of collected SEs is produced (Fig. .). If the bias voltage of the scintillator electrode switches off or if a negative bias is applied to the collector grid, only highly energetic BSEs can reach the scintillator, which leave the sample and head towards the detector. Due to geometric constraints, the signals from sample locations turned away from the detector do not contribute to the imaging; the resulting BSE image shows shadow phenomena. This mode is generally used as a simple method for BSE imaging. Another BSE detector with a higher detection efficiency is described below.

Fig. 4.10. Principle of the Everhardt–Thornley detector

4.3 The Instrumentation of SEM

99

The scintillator material consists of either a layer of fluorescent powder (placed on optically transparent platelets) or an yttrium-aluminium-garnet (YAG) crystal. The first type can degrade due to prolonged bombardment with high-energy electrons and hence need to be replaced at regular intervals. Other Detectors (Including Solid State Backscattered Electron Detectors) In addition to the type of detector discussed above, there are a large number of commercially available detectors that can be bought as complementary equipment for SEM. However, most of these are rarely used in polymer research. Only a few of them have found application in special electron microscopic tests, such as BSE detectors based on the electron voltaic effect in a solid state detector, and energy dispersive X-ray detectors, which will be discussed in Sect. ... The backscattered electron detector consists of two or four semiconductor diodes which are symmetrically arranged around the opening of the pole piece of the objective lens. In these diodes, the backscattered electrons generate electron-hole pairs in quantities that depend on the energy and the intensity of the electrons. The electron– hole pairs can be partially separated by the inner electric field at the p-n junction. By passing the p- and the n-type parts through an amplifier, an output signal can be obtained which is proportional to the backscattered electrons reaching the detector. By suitably coupling the signal outputs from different diodes, one can generate both topography-sensitive as well as material-specific signals (Sect. ..). 4.3.4 Signal Display and Magnification The signals obtained by the detectors are simultaneously displayed on a monitor. Data representation can be achieved through either analogue signal processing on a CRT (as in the older generations of SEMs) or by digitisation after displaying the data with the aid of a computer. When the first option is used, the images should be obtained via photographic techniques. The digital technique offers the chance to obtain the sample information directly, as a digital image which can be saved, modified and transferred instantly in electronic form. Both of the imaging techniques have one feature in common: the surface scanning and the representation of the surface information on the display take place simultaneously, such that each specimen point corresponds geometrically to the identical location on the recorded SEM image. The magnification in an SEM is the ratio of the lateral length of the image displayed or printed to that of the scanned area! This definition makes it clear that the magnification always depends on how the information is presented. For example, an SEM micrograph of  cm   cm with   magnification has double magnification as when the picture is presented in a  cm   cm format. Therefore, it is usual to introduce a scale bar to an image calibrated by the microscope, which automatically appears in the SEM images. Otherwise a scale bar (also called a μ-marker) must be manually marked onto a micrograph. The reasonable magnification of SEM (Mr ) is determined by the resolving power of the resolving power of the human eye (m, ca. . mm) and the resolving power of

100

4 Scanning Electron Microscopy (SEM)

the microscope (r). The latter depends on the signal used (e.g. SE, BSE, X-rays) and corresponds to the lateral diameter of the generation volume of the electron beam interaction product which contributes to the signal. Mr =

m . r

(.)

The above statement means that the operator must calculate the upper limit of magnification using Eq. . before beginning the work. For instance, to compare an SE image with X-ray mapping, it is important to be aware that the SE image can be constructed from more pixels than X-ray mapping because the effective probe diameter for SE is on the order of – nm, whereas the X-ray probe diameter is in the order of several microns (μm), and also depends strongly on the acceleration voltage. Nevertheless, it makes sense to work with so-called empty magnifications, e.g. during X-ray point analysis. If one works with very high magnifications (for instance  -fold conforms approximately to a point analysis), it is possible to observe the beam stability on the screen during the analysis.

4.4 Contrast Formation and Charging Effects 4.4.1 Secondary Electron Contrast As already discussed in Sect. .., PEs interact with the sample in an elastic or an inelastic manner. During the diffusion of the PEs, one or more of the processes discussed in the above section may take place. It is common to employ Monte Carlo simulation in order to obtain a statistical picture of the movement of the PEs. Figure . presents, for instance, typical path simulations for a light element (carbon) and a heavy metal (gold). For the lighter element and for high PE energies, the presence of a pear-shaped path distribution is typical. For heavy metals with lower PE energies, the distribution assumes a shape similar to that of a sectioned sphere. As a result, the penetration depth of the PEs for a lighter element is higher than that observed for a heavier one. SEs are generated during the diffusion of the PEs, but possess only a small amount of energy. Thus only the SEs produced by the surface layers can leave the surface and reach the detector. The SE efficiency and hence the signal-to-noise ratio of the image increases due to inelastic collisions close to the surface. The SE yield for elements with high atomic number Z is greater than for materials with low atomic numbers at identical PE energies, which means that the operator can reduce beam current at high Z. The lateral resolution power of SE imaging is affected by other factors as well as the beam diameter. Besides the dependence on the electron optics, there is also a dependence on the interaction of the PEs with the sample, because SEs will be emitted from a larger region of the specimen. In particular, for light elements (such as those present in plastics), a considerable number of secondary electrons are produced far from the location of PE incidence, which ultimately deteriorate the lateral resolution. The resolution can be improved in such cases by coating a thin film of

4.4 Contrast Formation and Charging Effects

101

Fig. 4.11a–d. Monte Carlo simulations of the PE paths for carbon (C) and gold (Au) at different PE energies (EP ): a C, EP = 5 keV; b C, EP = 25 keV; c Au, EP = 5 keV; d Au, EP = 25 keV

heavy metal onto the sample surface. Such a coating also contributes to reducing the surface charging (see Sect. ..). The improvement results from the corresponding interactions of the PEs with the heavy metal layer. At the same time, more SEs are produced near the primary beam zone, which improves the signal-noise ratio. The details of the SE emission determine the contrast in a SE image. Generally, good contrast is generated when the atomic number (and therefore the density) of the constituents differ significantly (see Fig. .). In polymeric materials, this kind of contrast (i.e. which depends on the atomic number) cannot be expected. The contrast formation in the SE mode is mainly determined by the local inclination of the sample surface with respect to the incident beam. This phenomenon, which is particularly apparent at surface edges (the socalled “edge effect”) can be explained by the correlation between the surface and the interaction volume of the PE, as represented in Fig. .. If the sample surface is not ideally flat, the interaction volume of the PE electron can pass through the side of the step, and from that location an additional number of secondary electrons can leave the sample surface, resulting in a higher SE signal. The maximum contrast enhancement occurs when the average penetration depth of the PE corresponds to the height of a step. Therefore, it can be deduced that the contrast in SEM imaging is determined not only by the atomic numbers of the elements but also the energy of the PEs selected for the SEM operation. Thus, to optimise the contrast, it is advisable not

102

4 Scanning Electron Microscopy (SEM)

Fig. 4.12. Atomic number contrast in SE mode of a polymeric concrete sample consisting of epoxy, quartz (SiO2 , large particles), calcite (CaCO3 , small particles) and spherical pores

Fig. 4.13. Contrast formation in secondary electron mode due to surface relief (right: SE micrograph of paraffin crystals, 12 keV)

only to work at a fixed acceleration voltage, but to vary it in order to match it to the sample. As well as a positive edge effect, a negative one can also occur; this is caused by the shadowing of SEs by the edge, and it results in a decreased SE signal.

4.4 Contrast Formation and Charging Effects

103

4.4.2 Contrast of Backscattered Electrons (Solid State Detector) The emission of the BSE can also be explained by the simulation scheme presented in Fig. .. Due to the strong interactions of the PEs and the resulting lower penetration depths for elements with higher atomic numbers, these elements offer a greater probability of BSE emission. On the other hand, only BSEs that reach the detector can contribute to the signal. The influences of the atomic numbers of the atoms within the region of interest and the surface topography on the signal detected by the individual segments (A and B) of a split BSE detector are illustrated in Fig. .. By forming a sum signal (A+B) and a difference signal (A−B) of the signals A and B measured by the detector segments, the influences of the kind of material and the surface topography can be distinguished. Compositional contrasts are produced by the sum signal (A+B), while the difference signal (A−B) results from shadowing effects of the surface relief and therefore provides an impression of the sample topography. Indeed, one should note that the efficiency of the BSEs is much lower than that of the SEs. Additionally, the noise levels of the detection and amplification systems significantly worsen the image quality when working with small beam currents in the BSE mode. In particular, the noise affects the difference signal (topography signal) of the split detector. Furthermore, it should be noted that BSEs from a surface region corresponding to the lateral extension of the interaction volume actually contribute to the BSE signal detected, while in the case of SE detection the signal mainly results from SEs emitted from the small area where the PEs meet the sample surface. 4.4.3 Charging Effect Both the secondary electron yield (ratio of the SEs emitted to the PEs that strike the sample) and the backscattering coefficient (the ratio of the BSEs emitted to the PEs that strike the sample) depend on the PE energy as well as the atomic number of the material and the angle of incidence of the PE []. Therefore, the charge supplied to the sample by incident PEs will generally differ from the charge release caused by the emission of SEs and BSEs. In most cases for bulk specimens, at PE energies of between several hundred eV and about  keV, the number of electrons that leave the sample exceeds the number of incident PEs, while at PE energies in the usual range (– keV) the opposite is true. Consequently, the charging of insulating samples occurs (Fig. .) apart from when the PE energy falls within a small range around  keV, where dynamic charge compensation takes place. For electrically conductive materials, the charge difference is compensated for by connecting the sample to the ground potential. To avoid the surface charging of insulating samples, their surfaces are usually coated with conducting layers of gold (Au) or carbon (C). However, it is important that the layer has sufficient contact with the specimen support held at ground potential.

104

4 Scanning Electron Microscopy (SEM)

Fig. 4.14. Demonstration of composition and topography contrast in BSE mode using two symmetric solid state detectors A and B (sample as in Fig. 4.12)

4.5 X-Ray Microanalysis

105

Fig. 4.15. SE image showing charging effect (porous UHMWPE reactor grains, 10 keV)

4.5 X-Ray Microanalysis 4.5.1 Physical Fundamentals of the Generation of X-Rays The imaging of surface structures (topography) or compositional variations on the surface is sometimes insufficient to fully characterise the specimen. If, for instance, inorganic particles are embedded in a polymer matrix, X-ray microanalysis [–] can be useful for accurately determining the nature of these particles. So-called “characteristic” X-rays are generated by inelastic interactions of incident primary electrons with the orbital electrons of specimen atoms, as already mentioned in Sect. ... Orbitals, which represent the spatial probability distributions of electrons, are defined by quantum numbers. According to Pauli’s exclusion principle, only one electron can have a given set of quantum numbers. The principal quantum number is the main influence on the binding energy and the distance between the nucleus and the orbital electron. The inner shells, designated K, L, M, etc., correspond to principal quantum numbers of , ,  . . . , respectively. The other quantum numbers have a relatively small effect on the energy and cause the shells (other than the K shell) to be split into subshells. If a localised electron is knocked out of an atom due to its interaction with a PE, the atom enters an excited high-energy state. At some later time, the empty electron orbital will be filled and the atom will relax, releasing the excess energy as a secondary effect either through the emission of a photon or alternatively through the

106

4 Scanning Electron Microscopy (SEM)

emission of an Auger electron (see Fig. .). If the vacant electron state is an outer state, then the excess energy will be small and the emitted photon will be in the visible wavelength range (cathodoluminescence). If, however, the vacant orbital is an inner one, the amount of energy released is greater, giving rise to the emission of an X-ray photon. The atomic states that are relevant to characteristic X-ray production can be represented as horizontal lines on an energy diagram, as shown in Fig. .. The energy of the emitted X-ray photon is equal to the difference between the two excited energy states (the original one of the vacant orbital and the final one of the orbital from which the electron jumps). However, it should be noted that not all transitions are allowed by the rules of quantum theory. The X-rays generated by transitions from any higher energy levels to lower K, L, or M shells are denoted K, L and M radiations, respectively. In the usual system of line nomenclature, the Greek letters α, β and γ refer to groups of lines with similar wavelengths, in order of decreasing intensity, while numerical subscripts distinguish the lines within each group, also in order of decreasing intensity. For instance, a transition from the L subshell to the K shell results in a Kα X-ray photon, while a K β X-ray photon is emitted by the transition from the M level to the K level. Further, it should be noted that the transition probability that leads to X-ray emission decreases with increasing distance between the energy levels taking part in the transition. As a result, for example, the intensity of the Kα X-ray peak is always higher than the corresponding K β emission peak.

Fig. 4.16. Energy levels and possible electron transitions for the emission of characteristic X-rays (after [15, 18])

4.5 X-Ray Microanalysis

107

Table 4.1. Characteristic X-ray peaks of elements occurring in polymeric materials [20] Element

K α peak

K β peak

C N O F Si Ca Au Ti

0.277 keV 0.392 keV 0.525 keV 0.677 keV 1.739 keV 3.690 keV

1.829 keV 4.012 keV

L α peak

L β peak

M α peak

0.341 keV 9.712 keV 10.267 keV

11.440 keV 12.211 keV

2.120 keV 2.268 keV

The energy of the characteristic radiation within a given series of lines varies monotonically with atomic number, i.e. the emitted radiation is element-specific. Therefore, for a qualitative chemical analysis, if the energy of a given K, L or M line is measured, the atomic number of the element that produces that line can be determined. Frequently observed characteristic X-ray emission peaks of elements commonly present in polymers are listed in Table .. Generally, the ionisation energy needed to excite an atom before the emission of a corresponding X-ray photon will take place exceeds the energy of the released photon. In practice, a rule of thumb is that optimal ionisation conditions are found when the energy of the incident PEs is about . times higher than that of the X-rays of interest. X-ray emission in SEM takes place in the whole interaction volume of PE with the sample (see Fig. .). Therefore, the lateral resolution of X-ray microanalysis is comparable to that of the BSE signals (see Sect. ..). As well as the characteristic (i.e. element-specific) X-rays generated, nonspecific radiation (called “bremsstrahlung” or X-ray continuum radiation) is emitted over the same energy range. The latter results from Coulomb interactions of PEs along their paths through inhomogeneous electric fields of sample atoms. Bremsstrahlung produces a continuous spectrum with the primary electron energy as the upper limit (this corresponds to the lowest value λ min in terms of wavelength; see Sect. ..). The contribution from bremsstrahlung must be removed from the measured X-ray spectrum to attain element-specific information. The superpositions of characteristic peaks and continuum radiation on the X-ray spectrum are shown in Fig. .. 4.5.2 X-Ray Microanalysis Techniques Two different experimental methods for X-ray microanalysis have been developed depending on the measuring technique applied: – Energy dispersive X-ray microanalysis (EDX) – Wavelength dispersive X-ray microanalysis (WDX).

108

4 Scanning Electron Microscopy (SEM)

Fig. 4.17. Scheme of a typical X-ray spectrum. Dotted line: X-ray continuum, solid lines: characteristic peaks

As the names suggest, in each case the emitted X-ray will be analysed either as a function of the energy of the emitted radiation (EDX) or its wavelength (WDX). Both of these techniques are discussed briefly in the following sections. 4.5.3 Detector for EDX Analysis EDX analysis [–] means energy dispersive spectroscopy of the X-rays emitted from a sample during electron irradiation. The detection technique resembles that of BSE solid state detector. X-rays penetrating the semiconductor detector are absorbed and generate electron–hole pairs. The formation of such a pair in a silicon semiconductor requires an energy of approximately . eV. Thus the number of pairs n e h formed by the total absorption of the energy Eph of one X-ray photon can be expressed by: ne h =

Eph . . eV

(.)

In order to determine the number of electron–hole pairs by measuring the corresponding current pulse, it is necessary to separate the pairs using an electric field in order to stop them from immediately recombining. To achieve this, an extended (Si(Li)) region in the initially n-doped silicon detector crystal is formed using drifted lithium. The charge carriers generated in this region have sufficient lifetimes due to the dominance of intrinsic p-i-n conduction. Furthermore, the electric field required

4.5 X-Ray Microanalysis

109

to separate electron–hole pairs in this region is produced by applying a voltage between the outer gold contacts of the detector in such a way that the p-n junction formed at the interface between the drifted lithium region and the adjacent n-doped region is reversibly biased (see Fig. .). The “separated” charges can be amplified in a field effect transistor (FET). To stabilise lithium within the detector crystal and suppress the thermal noise of the amplifier, the detector is usually cooled with liquid nitrogen. A very thin window transparent to X-rays separates the evacuated detection unit (inside a stainless steel probe) from the atmosphere within the SEM specimen chamber, and thus prevents the cooled detector from becoming contaminated. Windows of older systems consisted mainly of beryllium, whereas in modern systems a polymer-based ultrathin window (UTW) supported by a silicon grid is used. A sufficiently small time window is used for the basic detection cycle, during which all of the charges produced by this photon are integrated and processed to determine the energy of an individual detected X-ray photon. Using a multichannel analyser (MCA), the signal pulse is analysed based on its pulse height (which is proportional to the photon energy) and recorded as a count in the corresponding energy channel where all of these counts are accumulated. When the contents of the MCA channels are read out, an energy spectrum that shows how the accumulated counts of the channels (ordinate of the graph) depend on the corresponding photon energy (abscissa) is obtained, as demonstrated in Figure . for the spectrum of calcite.

Fig. 4.18. Scheme showing the working principle of an energy dispersive X-ray detector

110

4 Scanning Electron Microscopy (SEM)

In order to properly evaluate the resulting X-ray spectrum, different detector specific influences should be taken into account: . The window and the gold layer absorb low-energy X-rays. If a beryllium window is used, only a spectrum beginning with the Kα rays for sodium (Na) can be recorded. A polymer-based UTW enables the detection of photon energies down to the Kα radiation of boron (B); however, absorption effects are significant in this low-energy region. Below  keV, the transparency of UTW is energydependent, which might lead to “virtual” peak shifts and has to be taken into account for quantitative considerations. . Between the gold electrode and the p-i-n region is a dead layer in which the electron–hole pairs can immediately recombine. X-ray photons with penetration depths that do not exceed the thickness of the gold contact and the dead layer cannot contribute to the signal. . If the X-ray energy of a photon is high enough that it can penetrate into the n-conducting region of the detector, electron–hole pairs that form in this part of detector do not contribute to the signal either. Therefore, the energy determined is too low, resulting in a loss of intensity at the actual energy of the photon and increased intensity at lower (falsely determined) energy. . The value of . eV mentioned as the energy needed to produce an electron–hole pair in Si(Li) is only an average value. Actually, there are a wide range of energies, so different numbers of electron–hole pairs can be generated for the same X-ray energy. This causes the peaks in the MCA to broaden, and this deviation increases with increasing X-ray energy. Peak broadening can lead to the overlapping of neighbouring peaks. The measured full width at half maximum (FWHM) of the Kα radiation of manganese (Mn) is usually used as the energy resolution when evaluating the quality of a detector. Good Si(Li) detectors possess energy resolutions of smaller than  eV.

Fig. 4.19. EDX spectrum taken from a small particle (calcite, CaCO3 ) in Fig. 4.12

4.5 X-Ray Microanalysis

111

. X-ray absorption in the detector does not exclusively result in the formation of electron–hole pairs; it can also cause fluorescence emission from silicon (escape peak). Consequently, peak heights are strongly influenced by measuring conditions and will not directly yield quantitative information on local concentrations of elements. EDX detectors with take-off angles of between  and  are usually used. To increase the detection efficiency, the acceptance angle of the detector should be as high as possible, i.e. the detector must be placed very close to the sample. Furthermore, we must take into consideration that the X-rays to be detected should not be blocked by preceding surface structures on the sample. Therefore, the EDX analysis of a sample with a rough topography can be difficult. Compared to integral element analysis methods, the sensitivity of EDX analysis is relatively low; however, it provides the ability to carry out a microanalysis within an interaction volume that is in the micron or even submicron range. A limit of detectability (in terms of concentration) of about . wt% holds for elements of medium and high atomic numbers, while elements with low atomic numbers need a concentration of more than  wt% to be detected. 4.5.4 Quantitative EDX Analysis Quantitative EDX analysis [–] plays only a secondary role in polymer research. The reason for this is that the polymers are generally composed of the light elements carbon and hydrogen, with higher atomic number elements being present, if at all, only in very small amounts (generally below the detection limit; Sect. ..). Also, for polymers containing additives, the actual excitation volume of the additives is usually too small. The material in and around the excitation volume must be homogeneous. This requirement is usually not fulfilled by heterogeneous polymers containing relatively small particles. Additionally, a very flat surface is essential for a quantitative analysis. Therefore, only a short description of quantitative microanalysis will be presented here. As well as detector-specific properties, one should take also into account the X-ray emission in the excitation volume and the path of the photons in the material. Initially, a qualitative analysis must be carried out. As absorption and fluorescence take place in the detector itself, the silicon escape peak should be subtracted from the spectrum. The spectrum obtained after this step is subjected to peak separation (separation of overlapping peaks) and then the peaks are fitted with Gaussian functions. Then the X-ray continuum for all of the elements present is calculated and will be subtracted. Finally, the intensity under each Gaussian peak is integrated. This results in intensity values for each peak. Two methods can then be applied: K ratios or ZAF correction. K Ratios If the intensity values of the peaks of standard elements recorded under equal conditions are known, then quantitative analysis can be conducted relatively easily. To

112

4 Scanning Electron Microscopy (SEM)

do this, one constructs the so-called K ratio, which is the ratio of the measured peak intensity to the intensity of the standard sample peak. The result is, however, only a first approximation, because the real ratio corresponding to the absorption and fluorescence is not taken into account in this procedure. ZAF Correction Correction factors corresponding to the atomic number (Z), absorption (A) and fluorescence (F) are introduced in order to achieve an improved quantitative analysis. The Z term contains both the backscatter coefficients and the stopping power needed for the generation of an X-ray continuum. X-ray photons emitted in the interaction volume can be absorbed (A term) along their paths through the specimen. On the other hand, absorbed radiation can also initiate the emission of fluorescence radiation with lower energy (F term). Intensity values obtained and processed in this way can be used to conduct a quantitative analysis that considers all parameters. Comparisons with real standard samples (analysis with standards) or with virtual standards (standardless analysis) are made, whereas proportions of known elements that are not measurable can be calculated by a subtractive method. If the sample thickness is smaller than the excitation volume for the X-ray radiation (as is the case for ultrathin sections), this fact should also be taken into account. Manufacturers also deliver microanalysis systems with corresponding software that also take into account instrumental parameters for quantitative evaluations. During individual steps correction and calculation steps, it is also possible to carry out manipulations in an interactive way that minimises errors. In particular, when the specimen is not prepared as required for systematic studies (i.e. flat surface, homogeneous material, etc.), one should check the results for plausibility (e.g. the stoichiometry, see the discrepancy in quantitative results in Fig. . for SiO ). 4.5.5 X-Ray Mapping The SEM provides, beside the integral analysis discussed above (Sects. .. and ..), the ability to determine the elemental distribution (elemental mapping) over a sample surface. To do this, in an EDX system one or more peaks from an interesting element are selected from the spectrum by so-called regions of interest (ROI). While scanning the electron beam, it is possible to record the X-ray counts of the ROI correlating to points on the specimen surface. By accumulating X-ray counts per pixel, it is possible to create an elemental distribution image (X-ray mapping, see Fig. .). 4.5.6 Wavelength Dispersive X-Ray Microanalysis (WDX) WDX microanalysis [, ] was the original method used for elemental analysis by electron-induced X-ray emission. However, due to the time-consuming nature of this method, this technique is currently employed only when high spectral resolution is required (– eV), or element concentrations of less than .wt% need to be measured.

4.5 X-Ray Microanalysis

113

Fig. 4.20. EDX spectrum taken from a large particle (quartz, SiO2 ) in Fig. 4.12 and its quantitative (ZAF) results; C peak results from coating used to prevent charging

The relationship between the energy E and the wavelength λ of an X-ray photon is given by Planck’s relation: E = hν = h

c λ

(.)

where ν denotes the frequency of radiation, h = .  − Js is Planck’s constant, and c is the velocity of light in vacuum. For WDX analysis, the diffraction of X-rays by a crystal lattice is used to discriminate X-rays according to photon wavelength. Diffraction is an interference phenomenon, where intensity maxima only occur when the path differences between the waves diffracted in the same direction correspond to integer values n of the wavelength λ. Consequently, diffraction takes the form of a reflection of the incident beam at lattice planes, as illustrated in Fig. .. The relation between the glancing angle θ, the inter-planar distance d of the lattice planes and the wavelength λ is given by Bragg’s law: nλ = d sin θ .

(.)

Thus, the incident X-ray beam can be monochromatized by using an appropriately selected inter-planar distance d, and the wavelength of the monochromatized beam can be changed by varying the glancing angle. The WDX monochromator system comprises a crystal with a suitable value of d, an X-ray detector (a gas proportional counter) and a slit in front of it. The positions

114

4 Scanning Electron Microscopy (SEM)

Fig. 4.21. SE image, BSE image and X-ray mapping of C, O, Br, Os, Sb (cut surface for TEM preparation of flame-retarded HIPS with OsO4 staining)

of the sample, crystal and detector are arranged in such a way that the glancing angle of the radiation incident on the crystal and that of the radiation diffracted towards the detector slit are identical (see Fig. .). This condition is fulfilled only when the sample, the crystal and the detector are positioned on a circle called a Rowland circle. When the arrangement shown in Fig. . is used, the detector measures the X-ray intensity of the wavelength λ filtered according to Eq. . (usually for n = ) from the radiation striking the monochromator crystal, where the wavelength selected is

Fig. 4.22. Interference scheme for X-rays diffracted on crystal planes (Bragg’s law)

4.5 X-Ray Microanalysis

115

determined by the glancing angle used. Therefore, to record a wavelength-dispersive spectrum, the sample position is kept fixed and the glancing angle is varied by shifting both the monochromator crystal and the detector along the Rowland circle to the position needed. In order to make use of not only the central part but an extended region of the crystal in order to monochromatize the incident radiation, the crystal is concavely bent to a radius of curvature which is twice that of the Rowland circle and then ground in such a way that its surface has a radius of curvature equal to that of the Rowland circle. Such a spectrometer is known as Johnsson fully focussing spectrometer. When only one individual crystal (i.e. only one value of d) is used, the measurable wavelength range is limited for technical reasons. Therefore, WDX systems are often equipped with several monochromator crystals with different d values, as shown in Table .. The measuring conditions for WDX analysis require the exact positioning of the region of interest on the sample surface at the point of intersection of the electron optical axis with the Rowland circle. The adjustment is usually carried out with the help of an optical microscope attached to the SEM. Compared to EDX microanalysis, WDX analysis possesses a relatively poor signalto-noise ratio and requires a higher electron beam current. This method is therefore not suited to use as a routine analytical technique for polymers due to the risk of beam damage. When it is used for quantitative analysis, the effects described in Sect. .. should be taken into account.

Fig. 4.23. Scheme of a Johnsson fully focussing wavelength dispersive spectrometer: specimen, crystal and detector are arranged on the Rowland circle

116

4 Scanning Electron Microscopy (SEM)

Table 4.2. Different types of monochromator crystals commonly used in WDX spectrometers Crystal

2d spacing of specific planes used (nm)

Region of detectable wavelength (nm)

Minimal detectable atomic number

LiF α-Quartz Pentanerythritol (PET) Rubidium acid phthalate (RAP) Potassium acid phthalate (KAP) Pb stearate

0.40 0.67 0.87 2.61 2.66 10.4

0.08–0.4 0.12–0.6 0.15–0.8 0.5–2.4 0.5–2.4 2.0–9.0

19 15 13 8 8 5

4.6 Environmental Scanning Electron Microscope (ESEM™) 4.6.1 Low-Vacuum SEM and ESEM™ The environmental scanning electron microscope (abbreviated to ESEM) is a modified SEM that offers new applications and advantages over the conventional SEM, as described in more detail in Sects. .. and ... Used in this context, the term “environmental” refers to the possibility of imaging wet samples such as biological samples using the SEM. The ESEM was developed in the s. At that time, many SEM manufacturers introduced scanning microscopes that had low vacuums of about − mbar in their specimen chambers, which were achieved using a special pumping system. In these low-vacuum SEMs, charging effects can be strongly reduced and hence isolated samples can be imaged without coatings. Unfortunately, an Everhardt-Thornley detector (see Sect. ..) cannot be used in a low-vacuum SEM, as electrical breakdown at the high-voltage part of the detector cannot be avoided. Therefore, most producers of low-vacuum SEMs only have used BSEs to image the sample surface, which of course leads to limitations on the lateral resolution, as described in Sect. ... Danilatos and coworkers developed and patented [] a relatively simple SE detection system for a vacuum exceeding  mbar, which found application in a microscope called an ESEM []. The first ESEMs were constructed and commercialised by the company ElectroScan. Today, “ESEM” is a trademark of FEI. 4.6.2 Avoiding Charging As described in Sect. .., incident PEs usually cause negative charging of the surfaces of electrically nonconducting samples. Under low-vacuum conditions, a lot of residual gas molecules are ionised by the primary electron beam. The resulting positive ions are attracted by the negatively charged sample surface, and so the surface charge is compensated for by the impacting ions. Therefore, the surface charging of insulating samples is avoided or strongly reduced in a low-vacuum SEM. The presence of an additional electric field, as used in a “gaseous secondary electron detector”

4.6 Environmental Scanning Electron Microscope (ESEM)

117

(see Sect. ..) can further enhance this effect, because more ions are formed by cascade ionisation, which are then available for neutralisation of the surface charge. 4.6.3 The Wet Mode The ability of an ESEM to investigate wet samples is of great practical importance. Specimens do need not to be dried before the investigation, and they are not dried during the investigation either, as the pressure in the specimen chamber of the ESEM is comparable with the partial vapour pressure of water at room temperature. By using a differential pumping system [], which involves introducing some diaphragms to separate regions with different levels of vacuum along the electronoptical axis from one another, an ESEM can work with a relatively high pressure in the specimen chamber while a high vacuum or even an ultrahigh vacuum is utilised in the microscope column and the electron gun. The maximum pressure achieved in the specimen chamber depends on the diameter selected for the last diaphragm of the pressure system. A smaller one must be selected to maintain a high vacuum in the column at increased pressure in the specimen chamber. A small diaphragm, however, affects the field of vision, especially at lower magnifications and lower working distances. In the so-called “wet mode”, the previously evacuated sample chamber is vented to a pressure of some mbar of water vapour, so that, to a good approximation, only water molecules are present in the chamber. Depending on the microscope type, the maximum achievable pressure can range from less than  mbar to  mbar. Figure . shows the phase diagram of water close to its triple point. One can see that the transition from the liquid state to the gas phase at room temperature occurs at a pressure of  mbar. Therefore, if the water vapour pressure inside the sample chamber is  mbar and if the sample contains water, there will be a thermodynamic equilibrium. Under these conditions, the sample does not lose water by boiling and

Fig. 4.24. Scheme of the phase diagram of water in the range usable in ESEM

118

4 Scanning Electron Microscopy (SEM)

Fig. 4.25. ESEM (GSED) micrographs of a hydrated dental composite. Left: ambient conditions 7 mbar and 277 K. Right: after drying at lower pressure (3 mbar, 277 K)

a film of water is not deposited onto the sample surface. Upon increasing the pressure, the specimen will be covered with a film of water, while decreasing the pressure causes the sample to dry out. When working with an ESEM equipped with a FEG, one cannot create a pressure of  mbar, but one can shift the working conditions along the liquid/gas phase transition towards the lower pressure region by cooling the sample, e.g. by means of Peltier cooling. The wet mode is the standard working mode of the ESEM. Water vapour can be produced without difficulty using distilled water. The water in the atmosphere also forms enough ions to eliminate surface charging. Therefore, this mode is ideal for water-containing samples as well as for imaging nonconducting polymeric materials without coatings. It is also a highly advantageous approach to use for in situ investigations of polymers. Besides micromechanical tests, special investigations involving variations in temperature, pressure or the use of a particular gas atmosphere can be carried out using this technique (see Chap. , Fig. .). The reactions of the sample with gas molecules and the adhesion properties of its surface can also be comfortably studied using ESEM. Figure . shows the effect of the surrounding atmosphere on SE imaging in the case of a hydrated dental composite sample. When imaging in wet mode (with the surrounding atmosphere at close to the partial pressure of saturated water vapour), sample drying can be reduced and the formation of cracks is avoided (left). When imaging at lower pressures, vacuum water is removed from the sample, so that cracks appear at the interface to the glass ceramic particles (right). If it is impossible to use the water vapour pressure for practical reasons, the ESEM can also be operated under other gases. If, for example, nitrogen gas is used, charging effects can also be suppressed and SE imaging can be achieved using the GSED (see below).

4.6 Environmental Scanning Electron Microscope (ESEM)

119

4.6.4 The Gaseous Secondary Electron Detector (GSED) The patented GSED [] uses cascade ionisation of the residual gas molecules to amplify the secondary electron signal. Figure . shows the principle of this detector schematically. An isolated, positively biased electrode with a central hole, which the primary electron beam can pass through, is placed directly under the last diaphragm of the pumping system. If the operator intends to use a BSE detector at the same time, SE detection using a collection electrode that is not centrally arranged is also possible. Secondary electrons that leave the sample surface are accelerated towards the positively biased electrode and ionise residual gas molecules along their paths to the collector. The free electrons thus generated are also forced to travel towards the collector electrode and can trigger new ionisation events. These cascade processes cause the number of originally released SEs to be multiplied enormously, resulting in a measurable current at the collector electrode, which can also be amplified electronically. The final current signal is proportional to the number of SEs emitted per unit time, and it can therefore be used to image the sample. However, it should be noted that this SE signal is also influenced by the number of residual gas molecules available for ionisation (i.e. the pressure in the specimen chamber), the ionisation probability (i.e. the type of gas molecules present), the distance and the voltage between the sample and the collector electrode. A large working distance and a high pressure are favourable for high signal amplification. However, because of the widening of the primary electron beam due to collisions with residual gas molecules (skirt effect), highly resolved imaging (particularly at low PE energies) requires the use of a small working distance and/or a low pressure.

Fig. 4.26. Scheme showing the principle of the gaseous secondary electron detector (GSED)

120

4 Scanning Electron Microscopy (SEM)

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26.

Knoll M () Z Techn Physik (): Ardenne Mv (): Z Phys (-): Ardenne Mv () Z Techn Phys : Ardenne Mv (): Elektronen-Übermikroskopie (in German). Springer, Berlin Zworykin VK, Hiller J, Snyder RL () ASTM Bull : McMullan D () PhD thesis, University of Cambridge, Cambridge Smith KCA () PhD thesis, University of Cambridge, Cambridge Everhardt TE, Thornley RFM () J Sci Inst : Castaing R () Application des sondes électroniques à une méthode d’analyse ponctuelle chimique et cristallographique. PhD thesis, University of Paris, Paris Coslett VE, Duncumb P (). In: Sjöstrand FS, Rhodin J (eds) Proc Stockholm th Conf Electron Microsc. Almqvist and Wiksell, Stockholm, p  Danilatos GD () Micron Microsc Acta (): Danilatos GD () Scanning : Danilatos GD () In: Bailey GD (ed) Proc th Annual Meeting EMSA. San Francisco Press, San Francisco, CA, p  Danilatos GD () In: Bailey GD (ed.) Proc th Annual Meeting EMSA. San Francisco Press, San Francisco, CA, p  Reimer L () Scanning electron microscopy: Physics of image formation and microanalysis, nd edn. Springer, Berlin Lee RE () Scanning electron microscopy and X-ray microanalysis. PTR Prentice Hall, Englewood Cliffs, NJ Goldstein JI, Newbury DE, Echlin P, Joy DC, Fiori C, Lifshin E () Scanning electron microscopy and X-ray microanalysis. Plenum, New York Reed SJB () Electron microprobe analysis, nd edn. Cambridge Univ. Press, Cambridge Chandler JA () X-Ray microanalysis in the electron microscope. In: Glauert AM (ed) Practical methods in electron microscopy, vol , part II. Elsevier, Amsterdam Friel JJ () X-Ray and image analysis in electron microscopy. Princeton GammaTech Inc., Princeton, NJ Russ JC () Fundamentals of energy-dispersive X-ray analysis. Butterworths, London Heinrich KFJ, Newbury DE () Electron probe quantitation. Plenum, New York Scott VD, Love G, Reed SJB () Quantitative electron probe microanalysis, nd edn. Ellis Horwood, New York Mancuso JF, Maxwell WB, Danilatos GD () US Patent    Knowles WR, Schultz WG, Armstrong AE () US Patent    Danilatos GD, Lewis GC () US Patent   

5 Atomic Force Microscopy

Atomic force microscopy, also referred to as scanning force microscopy, is obviously not a technique based on electron microscopy. However, it has been common practice in the last few years to supplement electron microscopy with atomic force microscopy, as the latter enables the straightforward surface characterisation of polymers and provides additional insight into the structure and properties of homopolymers, blends and composites. Therefore, a brief survey of the fundamentals and relevant applications of this technique to polymer research is presented in this chapter. Because of its special importance for polymer investigations, tapping mode atomic force microscopy will be described in detail, while other modes of operation will be introduced more briefly.

5.1 Introduction In preceding chapters, different methods and techniques of electron microscopy that are employed for the study of polymeric materials were described. While atomic force microscopy (AFM), also referred to as scanning force microscopy (SFM), is obviously not a technique that is based on electron microscopy, it has been common practice in the last few years to supplement electron microscopy with AFM, as the latter enables the straightforward surface characterisation of polymers and provides additional insights into the structure and properties of homopolymers, blends and composites. Therefore, a brief survey of the fundamentals and relevant applications of this technique to polymer research is presented in this chapter. For a detailed account of the fundamentals and applications of AFM techniques to polymers, the readers should consult the more concise recent reviews of the topic [–]. AFM belongs to the family of scanning probe microscopes (SPM), in which solid surfaces are scanned by extremely sharp mechanical probes. In an SPM technique, highly localised tip–sample interactions are measured as a function of position. Different types of SPMs are based on different kinds of interactions; the major types of SPM include: the scanning tunnelling microscope (STM), which measures electronic tunnelling current; the atomic force microscope (AFM), which measures interaction forces; and the scanning near-field optical microscope (SNOM), which measures local optical properties by exploiting the evanescent field, i.e. near-field effects. These SPM techniques allow the characterisation of a wide range of properties

122

5 Atomic Force Microscopy

(structural, optical, mechanical, magnetic and electrical) of solid surfaces in different environments (vacuum, liquid or ambient conditions) and at different temperatures. Due to their ability to offer nanoscale resolution and versatile applicability, SPM techniques have proven to be indispensable part of modern nanoscience and technology. A major revolution in the advancement of surface analytical tools occurred when the first STM was introduced in  by Binnig and coworkers []. STM enabled the first ever three-dimensional imaging of solid surfaces with atomic resolution. Binnig and Rohrer were awarded the Nobel Prize in Physics in  in recognition of their outstanding contributions. STM is based on the electron tunnelling effect, which occurs when the distance between two conductors is close enough that the probability of charge transition via tunnelling through the potential barrier results in a measurable current. Generally, this is applicable for distances of smaller than  nm. The tunnelling current measured between the probe and the sample increases exponentially as the distance decreases. Therefore, it can be used as an ideal means for controlling the distance between the probe and the sample surface over the nanometre and subnanometre ranges. Obviously, STM imaging is limited to surfaces that are electrically conductive to some extent. Therefore, there was a need to develop an SPM that could be employed for both conducting and insulating surfaces. Indeed, the invention of the AFM by Binnig [], and its introduction by Binnig et al. [] in  enabled surface images of conductors and insulators to be obtained with atomic resolution by utilising very small interaction forces between the apex of the tip and the sample surface to control the distance between them. The most basic function of AFM is to provide high-resolution imaging of the surface relief of a specimen between lateral scales of a few nanometres to about hundred micrometres, as demonstrated by the examples in Fig. .. Figure .a presents the contact-mode AFM height image of isotactic polypropylene (iPP). The sample was prepared by annealing a polymer film cast from its xylene solution. One can see the typical surface texture of the polymer, where different spherulites are clearly separated at their boundaries. Figure .b presents a tapping-mode height image of an iPP surface area with a boundary between two spherulites, thus revealing the detailed morphology of α-type iPP with its cross-hatched texture (at the top of the image) and β-type iPP with its parallel arrangement of lamellae (at the bottom). In Fig. .c, a sample area of only  nm   nm is imaged to show the morphology of the polystyrene-block-polybutadiene-block-polystyrene block copolymer (SBS triblock copolymer) with symmetric polystyrene (PS) end blocks and the total PS volume fraction of .%. A Fourier filter was applied to the tapping mode height image to reduce the noise, and so the image, which is of a high quality, shows the periodic pattern of the dark appearing, cylindrical butadiene domains in the PS matrix. That the AFM can be employed to image the structures of crystal surfaces down to atomic resolution is illustrated by the contact-mode height image of freshly cleaved mica (Fig. .d). The periodic arrangement of the struc-

5.1 Introduction

123

Fig. 5.1a–d. AFM images of different samples illustrating the ability of the technique to image structural details at different length scales: a contact-mode height image showing the distribution of spherulites in an annealed, solution cast film of isotactic polypropylene (iPP); b tapping-mode height image of an iPP sample showing the interface region between a α-type spherulite (upper part of image) and a β-type spherulite (lower part of image); c tapping-mode height image of a symmetric SBS triblock copolymer after image processing by a Fourier filter, showing the periodic pattern of darkappearing cylindrical butadiene domains in the polystyrene matrix; d contact-mode height image of a freshly cleaved mica surface after image processing by a Fourier filter, showing the periodic pattern of the lattice

tures (with a period of about . nm) corresponds to the atomic lattice of the mica surface. By applying special investigation modes, AFM can additionally be used to measure normal and lateral forces, adhesion, friction, elastic/plastic mechanical properties (such as indentation hardness and modulus of elasticity), electrical and magnetic properties in relation to the local position on the sample surface.

124

5 Atomic Force Microscopy

5.2 Methodical and Instrumental Fundamentals Figure .a shows the main part of a commercial AFM (a “Dimension ” from Digital Instruments Inc., a subsidiary of Veeco Instruments Inc., Santa Barbara, CA, USA) that has been widely used for polymer investigations in an ambient environment, and the corresponding sketch (b) illustrates the principle of its operation. The instrumentation shown is typical of AFMs which are designed to be used for the investigation of large samples. The sample, here mounted on a motorised xy-stage, is not connected to the parts of the equipment necessary for the operation of the AFM, and the force sensor (a cantilever with a sharp tip at its free end) mounted on the cantilever holder is fixed at the piezo-driven xyz-scanner. While the sample is held in position (which can be adjusted via the motorised xy-stage), the tube scanner (made of piezoceramic material and referred to as the “PZT tube scanner”) scans the sharp probe in the x,y-directions over the surface in a raster pattern and controls the local distance between the tip and the sample surface via the very precise scanner motion in the z-direction. The vertical scanner motion (z-direction) is driven by a feedback loop that is linked to the sensed interaction force. Unlike large-sample AFMs, small-sample AFMs (such as the widely used “Multimode” from Digital Instruments Inc.) that are primarily meant for high-resolution atomic- or molecular-scale imaging of sample surfaces hold the position of the cantilever holder fixed, and the implemented probe and PZT tube xyz-scanner are adjustably positioned below the cantilever. The small and lightweight sample (generally no larger than  mm   mm) is mounted directly onto the scanner. During scanning, the cantilever base remains stationary and the sample is scanned under the sensing tip. The distance-dependent interaction forces between the tip and the sample surface provide the physical basis for AFM. The distance dependence is, however, more complicated than the monotonic function with a strong rate of decay of the tunnelling current utilised in STM. Due to the superposition of long-range attractive forces and short-range forces, which change with decreasing distance from being attractive to being repulsive, the total interaction force is nonmonotonic and thus does not share the helpful characteristics of the tunnelling current. Consequently, when the tip is brought in close to the sample surface, at first it senses an increasing attractive force, which reaches a maximum value and then decreases. This attraction finally changes to a repulsion that strongly increases with decreasing tip–sample distance. Interatomic forces with one or several atoms in contact are – or – pN, respectively []. Thus, atomic resolution with an AFM working in contact mode is only possible with a sharp tip positioned on a flexible cantilever at a net repulsive force of  pN or lower []. The key component in an AFM is the sensor used to measure the force on the tip due to its interaction with the specimen. AFM probes consist of a forcesensing cantilever with a very sharp tip integrated at the lower side of its free end. The apex of the tip has to be as sharp as possible. The fixed end of the cantilever meets a (usually) prismatic block (the cantilever base), which is large enough to be

5.2 Methodical and Instrumental Fundamentals

125

Fig. 5.2a,b. Main part of the “Dimension 3000”, a large-sample AFM (a), and principle of its operation (b)

126

5 Atomic Force Microscopy

handled with tweezers. A cantilever with a correspondingly low spring constant is needed to sense small forces (. nN or lower), but at the same time a high resonant frequency (about – kHz) is required in order to minimise the sensitivity of the AFM to vibrational noise from its surroundings, which is in the range of few Hz to  Hz. Commercially available cantilevers are batch-fabricated from silicon or silicon nitride by applying photolithographic and etching techniques. Rectangular single-crystal silicon cantilevers with a pyramidal tip are most commonly used. Tips with a radius of curvature of – nm are commonly available. Variations in the cantilever length (– μm) and thickness (– μm) result in cantilever spring constants of .– Nm− . Thus, a cantilever with an appropriate force constant for the special requirements of the specific operation mode performed is utilised. The force on the tip due to its interaction with the sample is sensed by detecting the deflection of the lever. Taking into account the spring constant of the lever, it is obvious that the cantilever deflection will be very small (displacements are smaller than . nm). Therefore, in the pioneering work of Binnig et al. [], the small cantilever deflection was sensed via the tunnelling current between the cantilever and a tip placed in the close vicinity of the rear side of the cantilever. A few other detection systems, including capacitance detection, piezoresistive detection and optical detection by means of optical interferometry, optical polarization, laser diode feedback and laser beam deflection, have also been used since then; these are reviewed in e.g. []. Here, only the optical beam deflection system will be described. This has a large working distance, is reliable and insensitive to distance changes, and is capable of measuring both normal forces (by sensing the bending of the cantilever) and lateral forces (by sensing the angular changes caused by cantilever torsion). Therefore, this is the detection method most commonly used in commercial SPMs. The sketch shown in Fig. .b illustrates the principle of its operation. The laser beam source is adjusted in such a way that the vertical path of the focussed laser beam strikes the rear side of the cantilever near the free end of the latter, where the beam is reflected. As the cantilever is tilted downward at about  with respect to the horizontal plane, the reflected beam is separated from the primary beam and it can be deflected by means of an adjustable mirror system so that it meets a split-diode photodetector with four quadrants (also called a position-sensitive detector) in an appropriate manner. Perfect adjustment, which is carried out in the detached position of the cantilever, results in vanishing values of the difference signals (A + B) – (C + D) and (A + C) – (B + D), where A, B, C and D are the photointensities as measured by the individual quadrants of the split photodetector. Prior to scanning the surface, the cantilever is changed from the detached position to the measuring position in close vicinity to the sample surface. In this position, the difference signal from the top and bottom photodiodes, i.e. (A + B) – (C + D), provides the signal which senses the normal interaction force via the corresponding deflection of the laser beam caused by the cantilever bending. If the scan is initiated with a preselected deflection signal (i.e. with a preselected interaction force), topographical features will cause corresponding local changes in the deflection sig-

5.3 Modes of Operation

127

nal. Only small sample areas with an extremely low roughness (e.g. atomically flat cleaved surfaces, as shown in Fig. .c) can be scanned to image the surface relief by using the deflection signal directly (“deflection image”). This mode of operation is commonly referred to as “constant height mode”. The use of this mode to scan sample areas of high roughness will inescapably result in the cantilever becoming damaged. Therefore, the so-called “constant force mode” is the one most commonly used for topographical imaging. In this mode of operation, the normal force applied is kept constant during scanning by means of a feedback circuit. The preselected deflection signal is used as the reference value (the so-called setpoint value) of the feedback loop. Topographical structures give rise to local changes in the measured deflection signal, which are interpreted as “error signals” by the input of the feedback loop. As a response, the system instantly compensates by changing the tip–sample distance by an appropriate amount. The output voltage of the feedback loop, which is proportional to the error signal, therefore controls the z-motion of the xyz-scanner. This output voltage signal is thus a linear response to the actual height variations on the sample surface. Therefore, when it has been calibrated by investigating a specimen with a known surface profile, it can be used as “height image” signal that shows the surface relief of the scanned area in a quantitative manner.

5.3 Modes of Operation AFM techniques can be divided up based on whether dynamic (i.e. the AFM makes use of an additional probe oscillation) or static operational modes of the AFM are used. Common “contact mode” AFM and also the imaging of lateral forces in a “lateral force microscope” (LFM) take advantage of static operation modes, whereas “noncontact” AFM and “tapping mode” techniques are categorised as dynamic operation modes. Because of its special importance for polymer investigations, tapping-mode atomic force microscopy (TMAFM) will be described in more detail in this section, while other modes of operation will be more briefly introduced. 5.3.1 Contact Mode In this mode, lateral scanning of the probe over the sample surface (as described in the previous section) is initiated when the vertical bending of the cantilever reaches the set-point value (chosen by the operator) that corresponds to a repulsive tip– sample interaction. To understand the reason for this assessment of the set-point value, it is useful to consider tip–sample forces in more detail using a so-called “force– distance curve”. When a sawtooth voltage is applied to the z-control of the scanner, the force–distance curve describes the interaction force at a particular x,y-position based on the cantilever deflection signal as a function of the position of the scanner in the z-direction over a cycle. This cycle includes the tip’s approach to the sample

128

5 Atomic Force Microscopy

Fig. 5.3. Typical force–distance curve. A1 and A2 are the start and end points of the cycle, respectively; A1  B approach; B  C1 pull down, contact at C1 ; C1  D, D  C2 indentation region; C2  E adhesion region, E  F disengagement; F  A1 end of the cycle

surface and its retreat from it. Figure . shows the various features of the curve. The force measurement starts and ends with the sample being far away from the tip, at the rest position of the cantilever (points A and A ). As the extension of the z-piezo increases, the tip approaches the sample surface. As long as the tip is far from the sample surface (A  B), the cantilever shows no deflection, as indicated by the flat portion of the curve. However, as the tip approaches the sample to within a few nanometres (point B), the tip experiences an attractive force which causes the cantilever to bend downwards. If the attractive force exceeds the withdrawing force of the cantilever, the cantilever is pulled towards the sample and contact occurs at point C of the plot. This will generally take place when a very flexible cantilever with a low spring constant (required for high sensitivity) is used. With the tip in contact, further extension of the z-piezo causes the cantilever to bend upwards, reflecting repulsive tip–sample forces. This is represented by the sloped portion of the curve (C  D, D  C ). As the z-piezo retracts, the tip is stuck to the sample by the capillary force, i.e. it goes beyond the zero deflection (flat) line into the adhesive region of the graph (C  E). The action of the capillary force results in a hysteresis of the force–distance curve. Capillary forces are mainly caused by a layer of liquid contamination that is present on samples in air. Therefore, a pull-off force (the difference in force between E and F) is needed to disengage the tip at point E of the graph. The pull-off force can be used as a measure of adhesion between the tip and the sample. Therefore, in the so-called “force volume mode”, force–distance curves are collected at a large number of points in the area of interest, and the data at any particular force level is presented as a map []. When force data at strong repulsive force levels are used, surface maps of mechanical properties can be obtained, and maps of adhesive properties of the sample are provided by force data in the disengagement region. Similarly, the topography, local stiffness and adhesion can be simul-

5.3 Modes of Operation

129

taneously mapped by the “pulsed force mode (PFM)” []. In this mode, a complete force–distance cycle is carried out during scanning at a repetition rate of between  Hz and  kHz through sinusoidal modulation of the z-piezo. The special PFM electronics used avoid the need to completely digitise the curve and extract only the important features. The slope in the repulsive region is analysed for the local stiffness. Local adhesion is deduced from the value of the critical force where the tip and sample contact are detached, and the topography is obtained from the feedback control. If scanning is carried out at set-point deflection, which corresponds to the attractive force near the pull-off point, imaging can be achieved with the lowest force for a given probe. However, stable imaging under these conditions is difficult due to possible tip disengagement, and the capillary effect limits the minimal force. The unwanted influence of the capillary force can be avoided by placing the sample and the measuring probe under liquid. Applications of AFM to biological systems often take advantage of this aspect. However, imaging under liquid has its own limitations. Therefore, contact-mode AFM is usually carried out with at set-point force that corresponds to a small repulsive interaction represented by the sloped portion of the force–distance curve near to the points C , C . Topographical images with a vertical resolution of less than . nm (as low as . nm) and a lateral resolution of about . nm have been obtained with a SFM operated in the contact mode []. Although high-resolution contact-mode AFM is carried out at a normal force that is as small as possible, such a small repulsive force is within the range of chemical bond energies and is sufficient to wipe away atoms. This fact explains why defect-free atomic resolution has mostly been observed with AFMs working in contact mode. Therefore, measurements utilising attractive interactions in the (dynamic) noncontact imaging mode may be desirable for imaging with atomic resolution. Normal and lateral forces between the tip and sample in standard contact-mode AFM are of tens and hundreds of nanonewtons. On the one hand, it is important to take into account that those forces cause damage to the surfaces of soft samples during such investigations. Therefore, its applicability to polymers may be limited. On the other hand, a tip with a normal load acts as a small indenter on the surface, and so its slide at a defined velocity over the sample surface during scanning is influenced by friction effects. As the lateral tip–sample forces can be used as a measure of friction, variations in them are recorded in a lateral force microscope (LFM) or friction force microscope (FFM) by sensing angular changes of the cantilever via the difference signal (A + C) − (B + D) of the left hand and right hand sets of quadrants of the slit photodetector. In the so-called “friction mode”, scanning is preferably carried out in such a way that the scan lines are orthogonal to the long axis of the cantilever beam in order to increase the torsional signal. Provided that the components have widely separated mechanical properties, the measurement of frictional force can be used to perform compositional mapping of heterogeneous polymers such as blends and composites. However, sufficient smoothness of the sample surface is a prerequisite for the meaningful FFM imaging of heterogeneous polymers.

130

5 Atomic Force Microscopy

5.3.2 Force Modulation Mode Like the pulsed forced mode, the “force modulation mode” was introduced to extend contact-mode investigations by simultaneously mapping mechanical properties. For this reason, the probe or sample assembly is scanned with a small vertical (z-direction) oscillation (modulation) that is significantly faster than the scan rate. The force on the sample is modulated about the set-point scanning force such that the average force on the sample is equivalent to that in simple contact mode. Early designs added a modulation signal to the z-piezo of the scanner to induce the vertical oscillation []. In second-generation systems for force modulation, the probe is made to oscillate vertically by means of an additional piezo-actuator at its resonant frequency (– kHz). This actuator is positioned at the cantilever base, i.e. at the fixed end of the cantilever. Therefore, when scanning in contact mode, the cantilever and its base also moves up and down with a small modulation amplitude that is induced by the piezo-actuator. The photodetector sensing the cantilever deflection collects force modulation data (fast-changing amplitude signal of deflection) and topographical information (slow-changing averaged signal of deflection) simultaneously. The amplitude response depends on the local stiffness of the sample, as the tip bounces from a stiff region on the sample and the amplitude is large, whereas the deflection amplitude is small at a soft sample region due to possible tip indentation. Therefore, the local heterogeneity in the mechanical properties of multicomponent polymers can be imaged by the amplitude signal of the force modulation technique, thus allowing surface compositional mapping. Figure . shows height and amplitude images of “core–shell” rubber particles of modified poly(methyl metacrylate) (PMMA) collected during force modulation

Fig. 5.4. AFM height (a) and amplitude (b) images for rubber-toughened poly(methyl-methacrylate) recorded in the force modulation mode; note how the soft shells of the dispersed particles appear brighter in the amplitude image

5.3 Modes of Operation

131

imaging. The particular method of force modulation used for this investigation was slightly different from the one described above. Utilising special scanning modes (socalled “interleaved scanning mode” and “lift mode”) controlled by the Nanoscope IIIa controller from Digital Instrument Inc., each scan line was scanned twice. The surface topography along the scanned line was determined during the first (“main”) scan. Contact-mode AFM or alternatively TMAFM can be used to do this. The measured surface relief was stored as a “height” signal pixel per pixel along the scanned line. The second (“interleaved”) scan of the line was carried out under force modulation conditions. Based on the stored surface relief data, the tip approached the sample to an operator-adjustable value (corresponding to a repulsive set-point force) and additionally oscillated in the z-direction with a small amplitude, which was also adjustable. The amplitude of the response was collected as described above; however, due to the electronics of the controller, the brighter the image structure in the “amplitude” image the smaller the measured amplitude. Therefore, in Fig. .b the soft rubber shells of the particles appear brighter than their stiff surroundings (the hard cores of the particles and the PMMA matrix) due to tip indentation and a correspondingly smaller deflection amplitude. In spite of the ability of the force modulation technique to image heterogeneous polymers based on local differences in mechanical properties, the technique is sample-unfriendly because it is still possible to damage the surface. 5.3.3 Dynamic Operational Modes In the dynamic mode of operation, the cantilever is excited such that it vibrates at or near its resonant frequency. Under the influence of tip–sample forces, the resonant frequency and consequently also the amplitude and the phase of the cantilever vibration will change and serve as measurement parameters. This is the basis for socalled “dynamic” AFM, which has been reviewed recently, e.g. in []. Two modes of operation dominate the application of dynamic AFM: amplitude modulation (AM) and frequency modulation (FM). In the amplitude modulation mode, the actuator is driven by a fixed amplitude at a fixed frequency f . When the tip approaches the sample surface, the interaction force between the tip and the sample causes changes in the amplitude and in the phase of the vibrated cantilever. These changes can be used as the feedback signal. While the AM mode was initially introduced as a noncontact mode, it was subsequently successfully used at a closer distance range that involved encountering attractive and repulsive interaction forces during each vibration cycle. This mode of operation, called the “tapping mode” (TM) or the “intermittent mode”, has been widely used for AFM investigations of polymers under ambient conditions. Therefore, it will be considered in more detail in the next section. As the change in amplitude occurs on a timescale τAM   Q f− in AM mode, the scan must be performed at a correspondingly slow speed if the investigation is carried out with a high quality factor Q of the vibration in order to reduce noise. In the FM mode, introduced by Albrecht et al. for analysing magnetic forces [], a cantilever with a high Q factor is driven to oscillate at its eigenfrequency by performing positive feedback with an electronic circuit that keeps the amplitude of the

132

5 Atomic Force Microscopy

cantilever vibration constant. The change in the eigenfrequency due to tip–sample interaction occurs on a timescale τFM  f − , and so the benefits of a high Q factor and high-speed scanning are combined in this mode of operation. This combination results in improved resolution, and when applied to noncontact AFM performed under ultrahigh vacuum conditions with a very small minimal distance between the tip and the sample, it provides the means to achieve true atomic resolution in AFM investigations. Although noncontact AFM in the FM mode is one of the most powerful AFM methods due to its enormous potential for modern nanoscience and technology, its applications to polymer research have been rather limited. The need for vacuum conditions is undoubtedly one reason that this method is rarely demanded by polymer researchers, particularly in the case of routine applications. Therefore, a more detailed description of this technique is beyond the scope of this section, and interested readers should consult concise recent reviews on the topic, e.g. [, ]. 5.3.4 Tapping Mode The principle of TMAFM and the symbols used in the following discussion are illustrated in Fig. .. The cantilever with the probing tip is forced by a driven actuator at the cantilever base to oscillate with certain amplitude A  of free vibration that is typically at or near its resonant frequency f  . The cantilever is then brought close to the specimen and made to tap the surface with a certain reduced set-point amplitude Asp .

Fig. 5.5. Illustration of the principle of TMAFM

5.3 Modes of Operation

133

During each vibration cycle the tip taps the sample for only a very short time. Thus, lateral forces and surface damage are avoided as much as possible during scanning. Therefore, this gentle AFM mode is the preferred method for investigating soft materials like polymers. The probe–sample interaction also results in a shift of the resonant frequency and in a phase shift ΔΦ of the vibration with respect to that of the freely oscillating cantilever. The resonant frequency and the phase shift are sensitive measures of the forces acting on the probe. Attractive forces acting on the AFM probe cause a negative shift in its resonance frequency, while repulsive ones lead to a positive shift. In the usual TMAFM imaging scheme, the specimen surface is scanned at a set-point amplitude Asp that is kept constant by a feedback loop. During the scan the vertical displacements Δz needed to keep the amplitude constant are displayed as a “height” image and the locally varying phase shift ΔΦ is displayed as a “phase” image. In principle, the height image should reflect the sample topography, while phase images show morphological structures of heterogeneous polymers. However, unlike contact-mode AFM, where force measurement by means of the deflection signal is straightforward and hence the formation of contrast in the scanned area is well-defined, the tip–sample interaction in TMAFM is rather complex. Therefore, the contrasts of the height and phase images strongly depend on experimental conditions. Factors that significantly affect the height and phase images in TMAFM of multicomponent polymers are the cantilever force constant, the tip shape, the amplitude A of free vibration and, in particular, the set-point amplitude ratio rsp = Asp A , as described in e.g. [, –]. Results from systematic TMAFM studies of a flat polymer blend surface, where height and phase images were simultaneously recorded at varied set-point amplitude ratios (rsp = Asp A ranging from . to .), are shown in Fig. .. Using a Dimension  AFM and a commercial NCL silicon cantilever from Nanosensors with a resonant frequency of  kHz and a cantilever spring constant of about  Nm, the investigations were performed under ambient conditions by driving the cantilever with an amplitude of A   nm at its resonant frequency. The material used in this study was a blend consisting of  wt% high-density polyethylene (HDPE) and  wt% ethylene/-octene copolymer (EOC). Both of the components, HDPE E (density . gcm ) and EOC AFFINITY* EG  (density . gcm ), were commercial products from the Dow Chemical Company. Due to the phase separation of the components, the blend shows a matrix with a morphology resembling that of the pure EOC. The formation of regions with ordered crystalline lamellae, as found in the pure HDPE, is hindered in the blend; instead, heaps and bundles of crystalline lamellae are embedded in the matrix. Nearly the same specimen area is recorded in the series of TMAFM height (a,c,e) and phase (b,d,f) images at set-point ratios rsp of . (a,b), . (c,d) and . (e,f). Due to the variation of rsp , small changes in contrast are observed in the height images (a,c,e). Phase image (b), recorded at light tapping with rsp = ., shows no contrast. However, lamellar structures and matrix regions can be clearly distinguished in the phase images (d) and (f), recorded at harder tapping with set-point amplitude ratios of .

134

5 Atomic Force Microscopy

Fig. 5.6. TMAFM height (a,c,e) and phase (b,d,f) images of the same area of the HDPE/EOC blend recorded at three different amplitude set-point ratios rsp of 0.95 (a,b), 0.5 (c,d) and 0.1 (e,f). The contrast covers height variations in the 120 nm range (a,c,e) and phase shift variations in the 100 range (b,d,f)

5.3 Modes of Operation

135

and ., respectively. On the other hand, the contrast inversion illustrates the problems involved in interpreting the contrast of TMAFM images, as discussed in detail in e.g. [, , ]. To examine the dynamics of the tip–sample interaction that cause the image contrast, additional TMAFM experiments where the lateral position of the tip is fixed and the amplitude signal and the phase shift ΔΦ of the tapping cantilever are measured as a function of the varied tip–sample distance Δz can be carried out. These TMAFM experiments resemble those performed in order to record force–distance curves, which are important for contact-mode AFM. Figure .a shows amplitude and phase shift versus distance curves recorded for various regions of the blend sample: in a soft matrix area of the blend (dotted lines) and at a specimen site where a bundle of crystalline lamellae was located (solid lines). These curves generally show the following typical features. When the drop in amplitude is small, a negative phase shift indicates that the overall tip–sample interactions are attractive. As the amplitude drops further, repulsive interactions become dominant, as seen by the switch in the phase shift from negative to positive. Finally, the phase shift drops to zero when the amplitude is reduced to zero and the vibration has ceased. Taking into account the initial amplitude and the correlation of the phase shift ΔΦ at an arbitrary point on the abscissa Δz with the corresponding reduced amplitude of the amplitude versus distance curve, the ΔΦ versus Δz curves were transformed into the ΔΦ versus rsp plots, as shown in Fig. .b. The differences between the two plots presented in Fig. .b recorded for the lamellar and the matrix regions of the blend, respectively, directly correlate with the contrast of the phase images at different rsp values. At light tapping with rsp close to , both plots almost coincide, and so phase shift contrast is not observable under this condition, as shown by Fig. .b. In agreement with Fig. .d, the greatest difference between the plots in Fig. .b at about rsp = . causes a maximum phase shift contrast. Lamellae appear bright under this condition, as the corresponding curve in Fig. .b exceeds that of the matrix region. A crossover of both curves takes place at about rsp = ., where the phase shift contrast vanishes again. A further decrease in rsp results in a contrast inversion of the phase images, as shown in Fig. .f. The interpretation of contrast in height images is more complicated. It takes advantage of an evaluation of the tip indentation δ in soft materials. Due to indentation δ, amplitude versus distance curves recorded on polymer samples deviate from the straight line that describes the drop in amplitude when the sample is hard, such as a silicon wafer. In the latter case, the amplitude drops linearly with the vertical distance and vanishes completely when the sample surface coincides with the cantilever baseline. Figure .a shows the amplitude versus distance curves of Fig. .a together with the corresponding curve obtained for a silicon wafer (thin solid line denoted “Si”). As described in detail in, e.g., [, ], the indentation δ is estimated at an arbitrary point of the amplitude versus distance curve for the polymer by calculating the difference between it and the corresponding reduced amplitude of the hard silicon wafer. As the slope of the Si amplitude curve is  , the difference in either the vertical direction or in the horizontal direction can be

136

5 Atomic Force Microscopy

Fig. 5.7. a Amplitude versus distance Δz and phase shift ΔΦ versus distance Δz curves of lamellar (solid lines) and matrix (dotted lines) regions of the blend. b Phase shift ΔΦ versus rsp plots of lamellar (solid lines) and matrix (dotted lines) regions of the blend, as estimated from a transformation (see text) of the ΔΦ (Δz) of a. (Reprinted from [22] with the permission of Elsevier)

used to obtain δ, as illustrated in Fig. .a. The estimation of δ as the difference between the ordinate values of the amplitude curves for the polymer and Si corresponds to the experiment used to record amplitude–distance curves, where the drop in amplitude with Δz is weaker in soft materials, due to indentation, than it is for hard materials, where no indentation takes place. On the other hand, the indentation δ expressed as the difference in abscissa values measured at the same reduced amplitude in the two amplitude versus distance curves corresponds to the variations in δ monitored in TMAFM height images, where the set-point amplitude is kept constant by changing Δz via the feedback loop. Figure .a also shows the estimated indentation curves, where solid and dotted lines again correspond to lamellar and matrix regions of the blend, respectively. Using the procedure applied to phase shift versus distance curves, a useful transformation into plots of

Fig. 5.8. a Amplitude reference curve of a silicon wafer (SI), amplitude versus distance Δz curves of Fig. 5.7a and estimated indentation δ versus distance Δz curves of lamellar (solid lines) and matrix (dotted lines) regions of the blend. The δ(Δz) curves were estimated using the ordinate or abscissa differences marked by δ (see text). b Indentation δ versus rsp plots of lamellar (solid lines) and matrix (dotted lines) regions of the blend, as estimated from a transformation of the δ(Δz) of a. (Reprinted from [22] with the permission of Elsevier)

5.4 Typical and Special AFM Applications

137

δ versus rsp (as presented in Fig. .b) was carried out. At light tapping with rsp smaller than about ., both plots are similar, and so height images monitored under this condition, such as Fig. .a, show the actual surface topography. When harder tapping (rsp < .) is used, the indentations in the lamellar and matrix regions of the blend differ significantly, and so height images recorded under this condition (Fig. .c,e) show a superposition of topographical and morphological information. In agreement with results reported in [], these investigations reveal that the profile in the height image only presents the true surface topography when rsp is close to . This image regime is known as “light tapping”. However, a much harder tapping with rsp  . is necessary to observe maximum phase shift contrast between stiff and soft regions of the blend. Thus, when TMAFM is used in the usual manner a single scan cannot yield optimum topography and morphology information. Therefore, interleaved scanning was suggested in [] as a way to simultaneously record a height image at light tapping, with rsp close to  in the main scan, and a phase image with maximum contrast at correspondingly harder tapping in the interleaved scan. Further examples of quantitative contrast interpretations of TMAFM height and phase images by means of results of amplitude/phase versus distance curves have been reported for different blends [, , ] and block copolymers [–].

5.4 Typical and Special AFM Applications It was mentioned at the beginning of this chapter that the application of AFM complements electron microscopic investigations of polymeric materials. Comparable results can be achieved by applying TEM and TMAFM in the usual way, as revealed by pairs of corresponding micrographs for the following examples: () arrangement of lamellar structures in α- and β-spherulites of isotactic polypropylene (Fig. .); () lamellar morphology of a styrene/butadiene block copolymer (Fig. .); phase-separated HDPE/VLDPE blend with elastomeric particles embedded in a semicrystalline matrix (Fig. .); Kraton SEBS triblock copolymer with cylindrical polystyrene domains (Fig. .). The examples provided here also make it clear that sample preparation varies depending on the morphological structures that are to be visualised by TEM and TMAFM. TEM investigations usually take advantage of samples which have been selectively stained, e.g. by OsO or RuO as described in detail in Sect. ., while hard and soft structural entities directly cause image contrast in TMAFM phase images if the differences in local stiffness are sufficient. In addition to the routine application of the AFM to characterise the morphologies of polymeric materials, the method has been extended to study many specific polymer properties. In such cases, special equipment is often attached to the AFM. Thus, due to the advantage of being able to visualise the sample morphology without chemically treating the sample, AFM is preferably used to investigate the local deformation behaviour in the micron and nanometre ranges by means of in situ tensile tests in an AFM equipped with an attached tensile module [–]. The corresponding results for a phase-separated HDPE/VLDPE blend are demonstrated in

138

5 Atomic Force Microscopy

Figs. ., . and ., while Fig. . shows a series of TMAFM images of a carbon nanotube-filled ethylene/octene copolymer deformed step-by-step to different strains. A heating stage is another very helpful AFM attachment. Using this combination, Hobbs et al. [] succeeded in a series of experiments in which processes such as crystallisation, crystal thickening and crystal deformation were followed in situ and in real time, providing significant new insights into long-standing problems in polymer science. When this technique was used to investigate the crystallisation of polyethylene shish kebab crystals in real time with nanometre resolution, images of the extended chain backbone and the overgrowth and subsequent interdigitation of lamellae were obtained []. Due to the gentle surface scanning associated with TMAFM investigations, they are also useful for studying changes in local surface regions caused by sample treatment at high resolution. Thus, by eroding the specimen step-by-step and imaging the same surface area via TMAFM after each step, Magerle [] expanded the AFM technique from surface imaging to volume imaging. By combining the series of TMAFM images in a similar way to computed tomography, this kind of “nanotomography” was used to reconstruct the -D distributions of polystyrene and polybutadiene in a poly(styrene-block-butadiene-block-styrene) [] and the crystalline regions in an elastomeric polypropylene []. Utilising the results of the TMAFM investigations described in the previous section and demonstrated in Figs. .–., the same HDPE/EOC blend was used to study the structural changes induced by different surface treatments. By applying interleaved scanning, TMAFM investigations of exactly the same specimen area were carried out before and after several surface treatments in order to evaluate the influence of the surface treatment in a direct way. The investigated surface treatments include chemical etching with a permanganic etchant, electron beam irradiation by scanning the surface area of interest in an environmental scanning electron microscope (ESEM), and plasma etching in an oxygen atmosphere. The results of these investigations are demonstrated by the corresponding series of TMAFM micrographs shown in Figs. .–.. In each series, the first row of micrographs shows the original untreated surface area while corresponding micrographs of the following rows reveal the modification of the sample due to the surface treatment in the area of interest. For light tapping, the height images in the first column of each series show the actual surface topography of the sample, while at harder tapping the blend morphology is revealed in the TMAFM phase images in the second column due to changes in the local stiffness of the sample. Starting from a very flat sample surface (micrographs a, b), Fig. .a,c,e (height images) illustrate the formation of surface relief with raised crystalline lamellae due to etching with a permanganic etchant ( ml sulfuric acid +  ml water + . g potassium permanganate) for  s (micrographs c, d) and  s (micrographs e, f). A selectively etched surface can be a helpful alternative for visualising the morphology of heterogeneous polymers if local stiffness differences are insufficient to cause contrast in the TMAFM phase image.

5.4 Typical and Special AFM Applications

139

Fig. 5.9. Series of TMAFM height (a,c,e) and phase (b,d,f) images of the same HDPE/EOC blend region before (a,b) and after permanganic etching for 15 s (c,d) and 30 s (e,f)

140

5 Atomic Force Microscopy

Fig. 5.10. Series of TMAFM height (a,c) and phase (b,d) images of the same HDPE/EOC blend region before (a,b) and after (c,d) ESEM inspection; secondary electron image of the ESEM inspection (e)

5.4 Typical and Special AFM Applications

141

Fig. 5.11. Series of TMAFM height (a,c,e) and phase (b,d,f) images of the same HDPE/EOC blend region before (a,b) and after oxygen plasma treatment for 10 s (c,d) and 40 s (e,f)

142

5 Atomic Force Microscopy

In Fig. ., the radiation damage caused by the impact of a -keV electron beam of low intensity on a selected sample area while scanning at a magnification of   in an ESEM-FEG XL environmental scanning electron microscope (Philips Electron Optics) for about half a minute is highlighted by the TMAFM images taken before (a, b) and after (c, d) the ESEM inspection. The area of interest in the ESEM was scanned at . s per image and recorded as an AVI video consisting of  individual images of size    pixels. While the original sample did not show any surface structure, the second image of the video showed the formation of surface relief due to the appearance of corresponding image contrast (Fig. .e). The analysis of the individual images of the video, performed by measuring the distances between distinctive image structures, revealed a central shrinking of the scanned area. The relative changes in lateral distances reach values of about . in the fifth image and . in the last image of the video. The shrinking observed, which corresponds to the deep central cavity recorded by the TMAFM height image (c), may be caused by electron beam-induced chain scissions and the volatilisation of small chain segments. On the other hand, a comparison of TMAFM phase images recorded before (b) and after (d) the ESEM investigation reveals that the contrast between the crystalline lamellae and their surroundings vanishes upon electron beam illumination. This equalisation of local stiffness is clear evidence that crosslinking has occurred. The reduction of contrast in TMAFM phase images of the blend due to crosslinking was also observed in corresponding experiments where TMAFM was applied to evaluate the influence of irradiation with gamma-rays and the changes caused by argon plasma treatment. A useful sample treatment leads to the results presented in Fig. .. This series of micrographs shows the original blend area of interest and the minor changes caused by oxygen plasma. The treatment was carried out in a PLS P plasma reactor by means of microwave plasma source (ECR-MW;  GHz;  W) at a working pressure of .  − mbar and an oxygen flow rate of  cm min− . The effects of selective surface etching can be detected in the height image (e), where a few raised crystalline lamellae appear after surface treatment for  s. Due to cross-linking in the EOC component, a small decrease in contrast is observed in the corresponding TMAFM phase image (f). This affects the sample for only  s, during which time the selective surface etching is suppressed and surface structures remain nearly unchanged, according to a comparison of the height images (a) and (c). On the other hand, the duration of the treatment is sufficient to clean the surface, and so unlike the original surface (b), both crystalline lamellae and matrix appear with uniform contrast in the TMAFM phase image (d). This example illustrates that a short sample treatment in oxygen plasma is useful since it cleans the surface by pushing away contaminated surface layers.

References 1. 2. 3. 4.

Magonov SN, Whangbo M-H () Surface analysis with STEM and AFM. VCH, Weinheim Magonov SN, Reneker D () Ann Rev Mater Sci : Tsukruk VV () Rubber Chem Technol : Jandt KD () Mater Sci Eng R:

References

143

5. Magonov SN () Atomic force microscopy in analysis of polymers. In: Meyers RA (ed) Encyclopedia of analytical chemistry. Wiley, Chichester, UK, p  6. Binnig G, Rohrer H, Gerber C, Weibel E () Phys Rev Lett : 7. Binnig G () Atomic force microscope and method for imaging surfaces with atomic resolution. US Patent ,, 8. Binnig G, Quate C, Gerber C () Phys Rev Lett : 9. Bhushan B, Marti O () Scanning probe microscopy: Principle of operation, instrumentation, and probes. In: Bhushan B (ed) Springer handbook of nanotechnology. Springer, Berlin, p  10. Ohnesorge F, Binnig G () Science : 11. Rosa A, Hild S, Marti O () Meas Sci Technol : 12. Maivald P, Butt HJ, Gould SAC, Prater CB, Drake B, Gurley JA, Elings VB, Hansma PK () Nanotechnology : 13. Schirmeister A, Anczykowski B, Fuchs H () Dynamic force microscopy. In: Bhushan B (ed) Springer handbook of nanotechnology. Springer, Berlin, p  14. Albrecht TR, Grutter P, Horne HK, Rugar D () J Appl Phys : 15. Morita S, Giessibl FJ, Sugawara Y, Hosoi H, Mukasa K, Sasahara A, Onishi H () Noncontact atomic force microscopy and its related topics. In: Bhushan B (ed) Springer handbook of nanotechnology. Springer, Berlin, p  16. Giessibl FJ () Materials Today : 17. Brandsch R, Bar G, Whangbo M-H () Langmuir : 18. Bar G, Thomann Y, Brandsch R, Cantow H-J, Whangbo M-H () Langmuir : 19. Whangbo M-H, Bar G, Brandsch R () Appl Phys A :S 20. Bar G, Brandsch R, Whangbo M-H () Surf Sci Lett :L 21. Bar G, Ganter M, Brandsch R, Delineau L, Whangbo M-H () Langmuir : 22. Godehardt R, Lebek W, Adhikari R, Rosenthal R, Martin C, Frangov S, Michler GH () Eur Polym J : 23. Bar G, Delineau L, Brandsch R, Bruch M, Whangbo M-H () Appl Phys Lett : 24. Kopp-Marsaudon S, Leclere Ph, Dubourg F, Lazzaroni R, Aime JP () Langmuir: 25. Knoll A, Magerle R, Krausch G () Macromolecules : 26. Leclere Ph, Dubourg F, Kopp-Marsaudon S, Bredas JL, Lazzaroni R, Aime JP () Appl Surf Sci : 27. Godehardt R, Rudolph S, Lebek W, Goerlitz S, Adhikari R, Allert E, Giesemann J, Michler GH () J Macromol Sci Phys B: 28. Michler GH, Godehardt R () Cryst Res Technol  : 29. Godehardt R, Lebek W, Michler GH () Morphology and micro-mechanics of phase-separated polyethylene blends. In: Grellmann W, Seidler S (eds) Deformation and fracture behaviour of polymers. Springer, Berlin, p  30. Hobbs JK, Winkel AK, McMaster TJ, Humphris ADL, Baker AA, Blakely S, Aissaoui M, Miles MJ () Macomol Symp : 31. Hobbs JK, Humphris ADL, Miles MJ () Macromolecules : 32. Magerle R () Phys Rev Lett : 33. Rehse N, Marr S, Scherdel S, Magerle R () Adv Mater :

6 In Situ Microscopy

Each type of electron microscope is suitable not only for studying the morphologies of polymers but also for visualising changes in these materials under different and varying conditions. Such experimental tests are commonly known as in situ techniques. Beside the effects of electron irradiation and different ambient atmospheres, micromechanical in situ tests are discussed here in some detail. Some examples are presented after describing the technical equipment used for in situ investigations, including miniaturised deformation devices for common electron microscopes, which enable tensile tests of ultra- and semi-thin specimens to be performed at low or high temperatures. SEM and (particularly) ESEM are very effective ways to study deformation, crack propagation and fracture processes. The TEM and HVTEM techniques permit higher resolution to be achieved and can be used to characterise effects at the micro- as well as the nanoscale. AFM can be applied to monitor micromechanical deformation processes without the limiting factors associated with EM, such as vacuum and electron irradiation damage.

6.1 Overview The great diversity of polymeric materials reflects their enormous structural variety from the molecular up to the macroscopic levels, which ultimately determines their physical properties. Therefore, it is fundamentally important for the development of materials with improved properties (e.g. mechanical, thermal, electrical, etc.) to gain an understanding of the relationships between morphology and these properties under different application conditions. Each type of electron microscope is suitable for studying not only the morphologies of polymers but also the changes in these materials under different and varying conditions. Such experimental tests are commonly known as in situ techniques. The term “in situ” is Latin, and means “in its place”. There are – depending on the type of microscope and the space available in its specimen chamber – several types of in situ experiments that can be performed: – – – – –

mechanical deformation tests heating or cooling electron irradiation application of electric or magnetic fields application of different ambient atmospheres.

146

6 In Situ Microscopy

Fig. 6.1a,b. In situ experiment performed in an ESEM (at 5 kV) to investigate the influence of the surrounding gas atmosphere on cellulose fibres. a: in a dry atmosphere (1.9 Torr N2 ); b: after 1 min of swelling in an H2 O vapour atmosphere

The main application of the in situ microscopy of polymers is the investigation of micromechanical processes [], which is discussed in more detail in Chap. .. Heating and cooling are often coupled with mechanical tests in order to study low- or high-temperature mechanical mechanisms. Additionally, the melting and crystallisation of semicrystalline polymers can be followed in situ, especially using AFM because this avoids the need for any electron irradiation []. However, the effects caused by electron irradiation can actually aid the morphological analysis of polymers. One result of electron irradiation is mass loss, which can differ significantly from one polymer to another. In Chap. , Fig. . shows a PVC/SAN blend in which the mass loss of the PVC phase is much greater that for the SAN phase. The dark PVC phase at the beginning of electron irradiation becomes much brighter after intense irradiation. Therefore, this irradiation-induced change in contrast can help us to identify different polymer phases. Another example is the improved contrast of spherulites in semicrystalline polymers; irradiation-induced contrast enhancement like this is discussed in detail in Chap.  (see, e.g., Figs. .–.). Studies of the effects of electric or magnetic fields do not play a crucial role in polymer research, since most polymers are electrical insulators and are even nonmagnetic. The influences of different ambient atmospheres can be studied using a special environmental (gas or liquid) cell for high-voltage electron microscopes (see Chap. ), in the chamber of an environmental SEM (ESEM, see Chap. ) or in the AFM (see Chap. ). Figure . shows an example where the shape of a sample changes when moisture is introduced into the atmosphere. The cellulose fibres are shrunken and close together in a dry atmosphere (Fig. ., left) but they expand in a moist atmosphere (right). In the following, in situ techniques for studying micromechanical processes of deformation and failure using different microscopes are discussed.

6.2 Micromechanical In Situ Tests

147

6.2 Micromechanical In Situ Tests 6.2.1 Technical Equipment Over the last few decades, significant advances have been made in the visualisation of failure mechanisms in materials upon loading. Significant improvements in measuring methods, loading apparatus design, and sample preparation have resulted in the application of in situ tensile tests, where the material is deformed by applying a tensile force while its deformation is observed using a microscope. A number of in situ techniques for scanning electron microscopes (SEM), transmission electron microscopes (TEM) and atomic force microscopes (AFM) have been developed to date. One of the most widespread is the in situ tensile test performed within an electron microscope, which is used to take sequences of pictures of deformation processes in real time. This test not only provides us with information about the “macroscopic” properties of the material, but also a deeper understanding of its deformation behaviour at the micro- and nanoscopic scales. In situ tools allow the local morphology and the deformation in a material that is subjected to local mechanical stresses to be mapped out. Load/deformation fields in the vicinity of crack tips in materials can be used to characterise and validate the various localised fracture mechanics models. In situ tensile tests on materials provide dynamic rather than static observations of the nucleation and propagation of cracks upon the application of a load. There are several in situ techniques for the investigation of micromechanical deformation processes [, –]. The appropriate in situ technique and tensile stage to use depends on the geometry, thickness, stress state and electron beam sensitivity of the samples. The data and specifications for several commercial tensile stages are listed and summarised in Table ., and the corresponding apparatuses are demonstrated in Fig. .. The tensile stages for SEM, ESEM and AFM can deform relatively thick specimens as well as semi-thin sections. Only semi-thin or ultrathin samples can be used for tests in HVTEM and TEM. Table 6.1. Data for commercially available in situ deformation devices (see also Fig. 6.2)

Device

Manufacturer

SEM

ESEM / AFM

Tensile module B156

Tensile and Tensile stage compression module 1000 N Kammrath & JEOL Weiss GmbH 0.5 μm to 1 mm 100 nm to 5 μm

Oxford Instruments Sample thickness 0.5 μm to 0.5 mm Temperature −180  C to +200  C range Resolution 10 nm

Room temperature ESEM: 5 nm AFM: 1 nm

HEM

TEM Single tilt cooling holder Model 671 Gatan

100 nm to 500 nm Room temperature −180  C to to +200  C +120  C 1 nm 0.5 nm

148

6 In Situ Microscopy

Fig. 6.2. Commercially available in situ deformation devices (a) for SEM, (b) for ESEM and AFM, (c) for HVTEM, and (d) for TEM (see also Table 6.1)

6.2.2 In Situ Microscopy in (E)SEM One of the most applicable in situ microscopic techniques is performed in a SEM. The tensile device for a SEM is shown in Fig. .a. Using this tensile stage, one can readily study crack nucleation and propagation in detail by stretching and/or compressing the relatively thick samples inside a SEM. The preparation of samples is comparatively easy since a conventional microtome can be used. Figure . shows a typical example of an in situ deformation test where crazes are formed in a semi-thin section of PS []. However, in most cases this technique is difficult to apply to polymeric materials because of the presence of charging electrons during in situ tensile tests. To overcome this limitation, a promising new technique is available, called environmental SEM (ESEM, see Chap. ). ESEM allows the examination of specimens surrounded by a gaseous environment. This means that the specimen does not need to be coated with a conductive material. Many types of samples like greases, adhesives, liquids, foods, gels, and other biological spices can be examined using this technique. One relevant example of the use of this technique is shown in Fig. .. Semi-thin sections with a thickness of about  μm from a PE nanocomposite modified with spherical SiO filler particles (which have an average diameter in the range of  nm) were microtomed from the bulk at −  C using a Leica Ultracut E ultramicrotome. In this nanocomposite, the severe formation of oversized agglomerates is observed, which

6.2 Micromechanical In Situ Tests

149 Fig. 6.3. Crazes in a deformed PS specimen 10 μm thick (SEM micrograph, deformation direction vertical). (From [1], with the permission of Hanser)

Fig. 6.4. Deformation structure of an agglomerate in a PE composite with 7 wt% SiO2 (ESEM micrograph, deformation direction vertical)

is a phenomenon that is commonly observed when working with nanofillers. These agglomerates, in general, result in a degradation of mechanical properties because of the premature failure of the material during the early stages of deformation (see Chaps.  and ). In Fig. ., a large agglomerate is moderately elongated in the tensile direction and fibrils are formed inside the deformed agglomerate, mostly at its equatorial region [].

150

6 In Situ Microscopy

An in situ deformation test in SEM of the step-wise development of craze-like deformation bands in a HDPE composite filled with  wt% Al O particles is presented in Chap.  (see Fig. .). 6.2.3 In Situ Microscopy in (HV)TEM Due to the increasing demand for higher resolution and magnification in order to investigate micromechanical deformation processes, the in situ TEM technique is in great demand and is still undergoing development. In general, small-scale deformation stages can be used to characterise the strengths of micro- and nanoscale reinforcements, the adhesive bond strengths of small structural features, as well as the kinetics of crack initiation and propagation. A representative device for this technique is shown in Fig. .d, for which ultrathin or not too thick semi-thin sections are usually used to allow the transmission of electrons. In other words, the materials should be “electron transparent”, which strongly depends on the atomic weight of the material, the accelerating voltage of the microscope and the intended experimental conditions. The sample thicknesses typically required for this technique lie in the range – nm, which can be achieved using a well-equipped ultramicrotome. As an example, the deformation structure observed from an impact-modified SAN with PBA core-shell particles (with an average particle diameter of about  nm) is demonstrated in Fig. .. Specimens about . μm in thickness were sectioned from the bulk at −  C using a Leica Ultracut E ultramicrotome, and then strained within the TEM while recording the deformation structures using a low-dose technique [].

Fig. 6.5. Deformation structure of core–shell modifier particles in a SAN/PBA blend at room temperature (TEM micrograph, deformation direction vertical)

6.2 Micromechanical In Situ Tests

151

The deformation structure shows the formation of microvoids inside the modifier particles in combination with the fibrillation of the PBA shells while most of the cores persist during the tensile loading of the specimen []. Micromechanical properties at low and high temperatures can be investigated using straining devices with cooling and heating facilities. Such a cooling/heating/ straining holder (shown in Fig. .d) makes it possible to deform thin sections at temperatures of between −  C and +  C at constant strain rates. Figure . compares micrographs of thin sections of a SAN/PBA blend subjected to in situ tensile tests performed at –,  and   C and a constant strain rate of . s− using a straining device in a  kV TEM []. The micrographs show obvious changes in deformation behaviour with temperature. At −  C (Fig. .a), the main mechanisms are crazing of the SAN and fibrillation of the elastomeric shells of the PBA particles around their PMMA cores (compare Fig. .). The bright crazes show up clearly against the dark background of the undeformed matrix and the paler grey of the moderately stretched modifier particles. At   C (Fig. .b), above the Tg of the rubbery phase, nearly all of the particles cavitate with fibrillation of the PBA shell, as shown in Fig. .. At   C (Fig. .c), there is very little evidence of crazing, and the matrix deforms almost exclusively by shear yielding in a ductile manner. As mentioned above, the requirement for ultrathin sections for in situ tensile tests with conventional TEM restricts its applicability. The higher the electron energy used, the thicker the specimen that can be proved. As a consequence, high-voltage electron microscopy (HVTEM) permits samples with thicknesses on the order of – μm, which may reveal bulk-like behaviour, to be studied. The tensile stage for HVTEM is shown in Fig. .c. As a typical example of the use of this technique, Fig. . shows a sequence of three micrographs taken at low magnifications during an in situ deformation test of an .-μm-thick section of PS, where the initiation of crazes between artificial cracks is shown (Fig. .b). The crazes appear as bright lines that are up to  μm long and possess a lower density because of their internal structure of microvoids and fibrils. After a small amount of additional deformation, the largest crazes break in a very brittle manner (brittle fracture, Fig. .c). The fibrillated structure of the crazes is visible at larger magnifications in Chap.  (see Figs. . and .) and the crack propagation inside the crazes is shown in Fig. .. Another example is provided by a PP blend. This blend is composed of polypropylene (PP) as matrix and polyamide  (PA) as the inclusion, which is surrounded by maleic anhydride-grafted polystyrene-block-poly(ethene-co--butene)-block-polystyrene (SEBS-g-MA) as compatibilizer []. The .-μm-thick specimens were sectioned from the bulk at −  C using a Leica Ultracut E ultramicrotome, and then strained within HVTEM using an accelerating voltage of  MV. Figure . shows a typical deformation structure of this material. During the deformation process, the shell consisting of SEBS-g-MA is first stretched in the tensile direction. Once the strain of the shell has reached a certain critical value, fibrils form at the interface between the PA inclusion and the matrix. These fibrils appear at the polar regions of the modifier particles and are aligned in the direction of applied stress. This kind of micromechanism is attributed to the relatively strong phase adhesion between the PA inclusion and the matrix [].

152

6 In Situ Microscopy Fig. 6.6a–c. Deformation structures in a SAN/PBA blend at various temperatures and at a constant strain rate of 0.05 s−1 : a −20  C; b +23  C; c +60  C. (Cryoultramicrotomed thin sections, TEM micrographs, deformation direction vertical, see arrow; from [15], with the permission of Chapman & Hall)

6.2 Micromechanical In Situ Tests

153

Fig. 6.7. In situ deformation of a 0.5 μm thick PS section in HVTEM (a) with artificial cracks before straining; (b) after deformation of about 10% with crazes; (c) after crack propagation and brittle fracture (for deformation direction see arrow; from [1], with the permission of Hanser)

Fig. 6.8. Deformation structure of a PP/PA/SEBSg-MA blend (HVTEM micrograph, the arrow indicates the deformation direction)

154

6 In Situ Microscopy

Fig. 6.9. In situ deformation with crack propagation and crack stop effect in a thin section of HIPS. (HVTEM, deformation direction horizontal)

Figure . shows crack propagation in a thin section of HIPS during in situ deformation in HVTEM. In a deformed area with crazes, a crack runs from the top, through a craze and is stopped in the soft rubber particle in the centre of the picture (the micrograph in the middle). After increasing the load, the craze below the large rubber particle rupture (right micrograph) and the crack continues to propagate through the rubber particle and the craze. This illustrates how the crack stop mechanism of the rubber particles is a precondition for high toughness in rubbertoughened polymers (see Chapter ).

6.3 In Situ Microscopy in AFM Up to now, we have established that AFM is a powerful technique for investigating the topographies as well as surface morphologies of materials. The advantages of working with this technique compared with electron microscopy can be listed as follows: – no vacuum is necessary – no sample damage occurs due to electron irradiation during experiments – sample preparation is relatively easy, since pretreatments such as staining, etching, etc., are not required.

6.3 In Situ Microscopy in AFM

155

Fig. 6.10a–d. In situ deformation test performed in AFM in tapping mode: a,c height images and b,d phase signal images; a,b before deformation and c,d after deformation

It was also recently shown that AFM can be used to monitor micromechanical deformation processes in real time (e.g. in situ) upon the external loading of materials. As an example, a blend of high-density polyethylene/-hexene copolymer with a weight fraction of / is shown in Fig. .. Specimens of size  μm   mm   mm were prepared from the bulk material at room temperature with a Leica RM microtome, and annealed at   C in order to flatten the rough surfaces caused by microtoming. In situ deformation tests were carried out in an AFM (Digital Instruments, D  atomic force microscope equipped with a Nanoscope IIIa controller) using a special device (see Fig. .b). The micrographs in Fig. . illustrate typical results from the AFM in situ tensile test. On the left hand side, height signal images show the surface topography, and the sample’s morphology is visible in phase signal images on the right hand side. A comparison of the micrographs shows that the area of  μm  . μm between the dotted lines in the micrographs in the first row is almost

156

6 In Situ Microscopy

homogeneously strained to  μm   μm in the tensile stress direction (marked by an arrow). This corresponds to a mean strain of the area of about %. In addition, the strong deformation of the elastomeric particles is accompanied by a very heterogeneous deformation of the surrounding semicrystalline matrix, and the alignment of crystalline lamellae into the direction of applied load is directly observed as the stress increases []. An additional improvement in the contrast between the lamellae and the amorphous parts can be achieved by image processing, as illustrated in the images of Fig. .b,d and in Fig. .b,d. The ability to perform and observe in situ deformation tests at the same sample position at large magnification – as illustrated in Fig. . – is a great advantage of AFM. However, it is difficult to get an overview of the sample during deformation at low magnifications, which is easy to do in ESEM or HVTEM. In this case a combination of optical microscopy and AFM is helpful. A small sample is deformed in the straining holder of an optical microscope, yielding pictures of the deformed sample at different stages of the stress–elongation curve; see Fig. .. The utilisation of a sample with an evaporated pattern of silver makes it possible to observe changes in the local deformation using the distances between the silver dots; see Fig. .. The silver dots also allow exactly the same sample area to be found using the AFM at larger magnifications. Figure . shows a thin section of a soft, very low density polyethylene (VLDPE) filled with  wt% carbon nanotubes (CNT). Using the silver dots, the same place can be found as the elongation increases. The sequence of micrographs in Fig. . demonstrates the local deformation behaviour of the stiffer

Fig. 6.11. Demonstration of an in situ deformation test where pictures were taken of a deformed miniaturised sample at different stages of the stress–elongation curve

6.3 In Situ Microscopy in AFM

157

Fig. 6.12a–f. Optical images of a miniaturised sample with an evaporated pattern of silver, recorded during in situ deformation test

carbon nanotubes in the soft polymeric matrix. Some nanotubes that were originally perpendicular or oblique to the loading direction are twisted into the tension direction.

158

6 In Situ Microscopy

Fig. 6.13. Series of AFM images of a carbon nanotube-filled polyethylene sample deformed step-bystep to different strains (0%, 36%, 75%, 155%)

References 1. Michler GH () Kunststoff-Mikromechanik: Morphologie, Deformations- und Bruchmechanismen. Carl Hanser, München 2. Pearce R, Vancso GJ () J Polym Sci Polym Phys : 3. Michler GH () Appl Spectrosc Rev : 4. Michler GH () Phys Status Solidi A : 5. Michler GH () Trend Polym Sci : 6. Michler GH () In: Salamone JC (ed) Polymer materials encyclopedia, vol  (D–E). CRC Press, Boca Raton, FL 7. Michler GH () Polym Adv Technol : 8. Kim GM, Michler GH () Polymer :,  9. Michler GH () J Macromol Sci Phys B:

References

159

10. Michler GH () J Macromol Sci Phys B: 11. Michler GH () In: Michler GH, Baltá-Calleja FJ (eds) Mechanical properties of polymers based on nanostructure and morphology. Taylor & Francis, Boca Raton, FL, Ch , pp – 12. Kim GM, Lee DH () J Appl Polym Sci : 13. Starke JU, Godehardt R, Michler GH, Bucknall CB () J Mater Sci : 14. Kim GM, Michler GH, Rösch J, Mühlhaupt R () Acta Polym : 15. Michler GH, Godehardt R () Cryst Res Technol :

7 Image Processing and Image Analysis

Today, the use of computer-assisted methods for the analysis and evaluation of microscopic images is inevitable. The initial task of image processing is to enhance the quality of digital images for further analysis. This optimisation comprises the use of greyscale, contrast, shading correction, specific filtering methods (e.g. sharpness, high pass, low pass, etc.), as well as arithmetic operations (e.g. addition, multiplication, logic operation). This chapter also reviews methods that are used to quantitatively determine specific image information, such as relative composition, particle size, interparticle distance, intensity profile, etc. Special image analysis procedures for the determination of periodicities in micrographs (such as the orientations of structural details, long periods, domain thicknesses, etc.) involve the use of Fourier analysis. The chapter also describes the application of stereoscopic imaging to show the topography of the sample surface, e.g. in scanning electron microscopy or optical microscopy.

7.1 Overview The microstructures of modern materials are becoming more and more complex and nonuniform. Therefore, it is very important to evaluate microscopy images in detail, objectively and quantitatively. Today, the use of computer-assisted methods for this analysis and evaluation is inevitable. Image processing and analysis aims to modify an image (such as an electron micrograph) using specific hardware and software in such a way that the quality of the image is improved and the quantification of structural details in the image is enabled [–]. A digitised image, which can be obtained through various means, is a prerequisite for image processing. Modern transmission electron microscopes equipped with a CCD camera or image plates as well as scanning electron microscopes and atomic force microscopes allow the direct recording of digital images of the objects investigated [,,,]. In older electron microscopes the images were recorded on films or negatives. These analogue images can be converted into digital images using a digital camera or scanner. Generally speaking, a “picture” is a reproduction of a structure in an analogous form. A digital image is a picture that has been created or converted into a discrete

162

7 Image Processing and Image Analysis

Fig. 7.1. Dependence of the perceptibility of the structural details of an object on the number of pixels; the object is a tensile module for HVEM

form. The crucial factors associated with the quality of a digital image are the resolution of the image (i.e. the number of image points or pixels, see Fig. .) as well as the bit depth used for each pixel (e.g. every pixel of an eight-bit greyscale image corresponds to  points on the grey scale). The number of image points (pixels) that can be included in the image is determined by the technique used to take the photograph. The greyscale value is typically between  and  bits in length. Images can be saved in different formats; one can differentiate between pixel and vector formats as well as between compressed and uncompressed ones. In microscopy, the uncompressed pixel-based TIFF format is generally used, which contains all the image information required for further processing and analysis. A description of other conventional image formats, such as those used in photography or computer-aided design, is beyond the scope of this chapter.

7.2 Image Processing The quality of the recorded image depends on the parameters of the microscope (such as the beam intensity in TEM) and camera parameters (such as illumination time, contrast enhancement, image integration, etc.). Processing the digitised picture

7.3 Image Analysis

163

Fig. 7.2a,b. Example of the optimisation of greyscale values; the object is a HVEM tensile module: a original image; b optimised image

enhances the quality of the image (such as greyscale, contrast, etc.), and the resulting image is hence optimised for further analysis. This optimisation comprises various operations, such as shading correction (equalisation of an inhomogeneous image background), specific filtering (e.g. sharpness, high pass, low pass, etc.) as well as arithmetic operations (e.g. addition, multiplication, logic operation). The full range of values available for adjusting pixel intensity often remains unused. This means that images will appear weak in contrast, or an image will seem too bright or too dark. An example of the optimisation of a greyscale image is shown in Fig. .. Using various filter operations, it is possible to highlight or attenuate specific structural details. After applying a low-pass filter (e.g. a Gaussian filter) to an image it will appear less sharp, with the edges of the grey areas becoming blurred and noise diminishing. In contrast, a high-pass filter will makes the image appear sharper, highlighting the finer structural details (such as edges) and removing the homogeneous regions of an image (see Fig. .).

7.3 Image Analysis There is a proverb that is often used by microscopists to brag about the value of their work: “A picture says a thousand words”. However, from the point of view of modern information theory, one would ask how much information these thousand words (the picture) actually provide. It is more important to extract valuable information about the materials from the microscopy images. Microscopists have always extracted such information about the materials by interpreting micrographs. This form of qualitative study is subjective in many cases. In the industrial world, the microstructures of samples are often very complex and nonuniform, so a large number of images have

164

7 Image Processing and Image Analysis

Fig. 7.3a–c. Example of filter operations leading to changes in image details: a AFM phase image of a P(E-co-H) blend; b after low-pass filter; c after high-pass filter

Fig. 7.4. a TEM micrograph of a filled SAN/PPO blend; b determining the volume fractions of the constituents by assigning defined greyscale values to individual components

7.4 Fourier Transformation

165

to be analysed. The subsequent image analysis serves to quantitatively determine the specific information of interest contained by the processed images, such as the relative composition, particle size, interparticle distance, intensity profile, etc. Automatic evaluation of the image information is generally possible provided that the measured image information can be separated from the background (greyscale separation and binarization). However, many images (in particular TEM micrographs) only permit an interactive evaluation, i.e. where the image details are marked using a computer mouse. The fundamental possibilities offered by image processing for the quantitative evaluation of structural details are clarified by the following two examples. Figure . illustrates how one can determine the volume fraction of the components in a SAN/PPO polymer blend. By defining the greyscale regions of the image under consideration (which practically reflect the amounts of the respective constituents), and calibrating the image (i.e. based on the total number of pixels), one can easily calculate the relative compositions of different phases in multicomponent systems. In Fig. . a direct (automatic) determination of the diameter of PS latex particles is presented using a special image analysis algorithm. The hardest task in automatic image analysis is to separate the individual latex particles, as there is no clearly visible boundary between them. This special analysis was performed using the image processing and analysis software “Analysis” (from Soft Imaging System, Münster, Germany).

7.4 Fourier Transformation Special image analysis procedures for determining the periodicities in micrographs (such as the orientation of structures, long periods, domain thicknesses, etc.) involve the use of Fourier analysis, which is based on so-called “fast Fourier transformation (FFT)” [, , ]. During the Fourier transformation of an image, the periodic components of the image are determined and presented as a function of the frequency of corresponding periodicity. The evaluation of the Fourier image enables the preferred orientation direction of the structures in the image to be determined. By evaluating the reflexes of the Fourier image followed by inverse Fourier transformation, one can also quantify the size of the structural periodicity. Figure . presents the procedure for determining the lamellar periodicity in a β-iPP. In the Fourier image (see the greyscale profile in Fig. .b, the regular lamellar spacing is represented by the reflexes. The average lamellar spacing can be calculated directly by evaluating each peak (Fig. .b) of the reflexes. Using this technique, additional details can be revealed for long-period distributions. Figure .a shows an electron micrograph of a shish-kebab structure in a HDPE. Direct optical measurement of the long period of the oriented lamellae shown in the micrographs revealed values in the range of – nm. Laser light diffraction patterns (Fig. .b) from the negatives of the electron micrographs show two maxima. Fourier transformation of the electron micrograph (Fig. .d) gives two maxima for long periods, L =  nm and L =  nm. Long periods of the

166

7 Image Processing and Image Analysis

Fig. 7.5a–e. Automatic determination of the diameter of PS latex particles: a original image; b binarised image; c separation of particles using special software; d particle size analysis and assignment in a predefined category; e frequency distribution of particle size

same length were also detected by SAXS measurements []. This means that scattering techniques (laser light, SAXS) and Fourier transformation yields long periods between lamellae of around  nm with clear contrast (laser light) and electron density (SAXS), whereas direct inspection of the micrographs also yields smaller long periods between thinner, not so strongly stained lamellae (in the range of  nm).

7.5 Stereoscopic Imaging

167

Fig. 7.6a,b. Determination of lamellar spacing in a β-iPP: a TEM micrograph of a stained ultrathin section (PP lamellae white, amorphous parts dark); b Fourier image with greyscale profile (line plot)

By applying specific filters, it is possible to enhance (or even eliminate) different reflexes of the Fourier image so that the back-transformation results in a so-called “corrected” image. This procedure modifies the image information somewhat, but the images gain clarity, and present clearly noticeable structural details (see Fig. . with a PB/PS block copolymer as an example). Using Fourier transformation, it is possible to correct different kinds of errors in electron micrographs (such as preparation-induced striations of sections, folding, etc.) or to eliminate artefacts that do not belong to the morphology of the investigated material (see Fig. .).

7.5 Stereoscopic Imaging Stereoscopic imaging is particularly suited to the illustration of surfaces in scanning electron microscopy or optical microscopy for example. Any stereoscopic imaging is based on the concept that the human brain computes spatial information from the different images that an object creates in the retinas of the left and right eyes [, ]. The difference between the images in the left and right retinas is caused by the different observing angles of the eyes. This is why a human is capable of seeing in three dimensions and estimating spatial distances. A stereoscopic image pair consists of two images that display the same object before and after a tilt by a certain stereo angle ω. Stereo angles of about − are generally used. The optimum tilting angle depends on the microscopic magnification and the roughness of the sample surface. Special software creates stereo images, for example in certain colour combinations (e.g. red–green or red–blue), which produce a three-dimensional impression if viewed through coloured glasses.

168

7 Image Processing and Image Analysis

Fig. 7.7a–e. Determination of long period of HDPE: a TEM micrograph of shish-kebab structure; b laser light diffraction pattern from negatives of electron micrographs obtained at lower magnification; c microdensitometer curve of (b) between arrows; d FFT spectrum of (a); e greyscale value distribution of (d) between arrows

The heights of image points can be computed based on two stereo images with a defined stereo angle ω between them. Due to the tilt, individual points on the object are usually at different positions in the images of the stereoscopic pair. In the initial position, the tilt axis must be parallel to the vertical borders of the images. The object is then tilted by −ω and the first image is taken. Then the object is tilted by +ω from the initial position and the second image is obtained. The height h of any point resulting from its x-shift xsh between

7.5 Stereoscopic Imaging

169

Fig. 7.8a–d. Optimum presentation of hexagonal arrangement of PB cylinders in a PS/PB block copolymer: a original AFM image (PB cylinders appear dark); b FFT image of (a); c inverse-transformed image after filtering; d modified FFT image with lattice filter

the two images is: h=

xsh . sin ω

(.)

When the height is measured, one image from the stereoscopic pair will be used as the reference image and the other as the search image. The height of a point is calculated by considering the x-shift between the reference point and the corresponding search point. Usually the individual image points have different heights and so they are shifted to different degrees and the search image is more or less distorted. When a human observer compares the two image patterns, he will use the terms “the same pattern” or “nearly the same pattern” to describe the degree of congruence. The degree of coincidence or correlation between the reference pattern and the distorted

170

7 Image Processing and Image Analysis

Fig. 7.9a–d. Elimination of preparation (ultrathin sectioning)-induced striations from a TEM image of an ABS material due to Fourier transformation and filtering: a original image; b FFT image; c inversetransformed image after filtering; d filtered FFT image

and redetected pattern in the search image can be described by a mathematical magnitude that ranges from  to . The closer the correlation value is to , the better the congruence between both patterns. There is no generally valid relation between the correlation and the accuracy of the calculated height. If the reference pattern is extremely distorted due to the tilt, a bad correlation should be expected but the resulting height might still be correct. On the other hand, if there are no clear structures in the images, the correlation might be excellent but the height is completely wrong, because the reference pattern cannot be detected unambiguously. Three-dimensional electron tomography is a technique in which -D information on a specimen is obtained by tilting the specimen in a transmission electron microscope and acquiring projection images at many different viewing angles. Subsequent

References

171

reconstruction in the computer reveals spatial information on the specimen under consideration. As tomography is increasingly being used as a routine research technique in microscopy, it has become necessary to automate and streamline the entire process of -D reconstruction, from sample preparation to visualisation and analysis of the resulting -D data cubes [–]. When attempting to obtain high-quality -D sample data using electron tomography, there are several aspects that are of major importance, some of which are associated with the TEM during the acquisition of the data, while others come into play during the subsequent reconstruction process (see also Chap. ).

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.

19. 20. 21. 22.

Castleman KR () Digital image processing. Prentice-Hall, Upper Saddle River, NJ Sawyer LC, Grubb DT () Polymer microscopy, nd edn. Chapman and Hall, London Kirkland EJ () Advanced computing in electron microscopy. Springer, New York Pitas I () Digital image processing algorithms and applications. Wiley, New York Rourdeyhimi B () Imaging and image analysis applications for plastics. William Andrew, Norwich, UK Zhang XF () Progress in transmission electron microscopy . Springer, Berlin Russ JC () The image processing handbook. CRC Press, Boca Raton, FL Yamanaka CR, Zhigang RL, Li L, Khigang RL () Industrial application of electron microscopy. Marcel Dekker, New York Jahne B, Jahne J () Practical handbook on image processing for scientific and technical applications. CRC Press, Boca Raton, FL Jähne B () Digital image processing: Concepts, algorithms, and scientific application. Springer, Berlin Umbaugh SE () Computer imaging: Digital image analysis and processing. Taylor & Francis, Boca Raton, FL Hader DP () Image analysis. CRC Press, Boca Raton, FL De Graef M () Introduction to conventional transmission electron microscopy. Cambridge University Press, Cambridge Graham D, Barrett AN () Knowledge-based image processing systems. Springer, Berlin Wojnar L, Wojnar W () Image analysis: Applications in materials engineering. CRC Press, Boca Raton, FL Michler GH, Naumann I, Baltá Calleja FJ, Ania F () Acta Polym : Russ JC, Dehoff RT () Practical stereology. Kluwer Academic/Plenum, New York Staniforth M, Goldstein J, Newbury DE, Lyman CE, Echlin P, Lifshin E, Sawyer LC Michael JR, Joy DC () Scanning electron microscopy and X-ray microanalysis. Kluwer Academic/Plenum, New York Frank J () Electron tomography: Three-dimensional imaging with the transmission electron microscope. Springer, New York Yaroslavsky L () Digital holography and digital image processing: Principles, methods, algorithms. Kluwer Academic, Boston, MA Schoenmakers RHM, Perquin RA, Fliervoet TF, Voorhout W, Schirmacher H () Microsc Anal : Fliervoet T () Imag Microsc :

Part II

Preparation Techniques

8 Problems Associated with the Electron Microscopy of Polymers

This introductory chapter to Part II – which deals with preparation techniques – summarises the main problems associated with the investigation of polymers by electron microscopy. The irradiation sensitivity of polymers can be reduced by taking precautions with the instrumentation and the manner of operation. It should also be noted that the irradiation sensitivity of polymers can be utilised for special contrast development effects. The problem of the low contrast between structural details of polymers can be overcome through the use of chemical staining, physical effects or surface etching. The third problem is the preparation of ultrathin specimens from bulk polymers, which is successfully solved by applying (cryo)ultramicrotomy. Some additional methods are mentioned, and the applicabilities of various methods for studying the morphologies of several classes of polymers are summarized. All of these methods are described in detail in subsequent chapters in Part II.

8.1 Overview The polymers that are studied by electron microscopy can take various shapes, forms and sizes. Examples include powders obtained directly from macromolecular synthesis and granules from compaction or a first extrusion step, and injection and extrusion-moulded test parts in the centimetre range or large pieces for different applications. In general, direct investigations of polymers by electron microscopy involve three problems: . The usual preparation techniques applied to inorganic samples, particularly for TEM investigations, cannot be applied, and the preparation of ultrathin specimens from bulk polymers is often difficult . Polymers (since they are organic substances) are particularly sensitive to electron beam irradiation . The contrast between structural details is often very low because polymers usually consist of the same light elements (C, H, O, and others) that interact only weakly with the electron beam (see Sects. . and ..). Several preparation and investigation techniques have been developed to overcome these difficulties. These are reviewed in this chapter, and the main steps are discussed in detail in subsequent chapters in Part II.

176

8 Problems Associated with the Electron Microscopy of Polymers

8.2 Irradiation Sensitivity of Polymers The effects of irradiation on polymers have been discussed in many papers and several reviews (e.g. [–]). The primary effects of the interaction of electrons with organic matter are inelastic scattering processes, which yield ionisation and break chemical bonds. Secondary effects are mainly chain scission or crosslinking, mass loss, fading of crystallinity, heat generation, and charging-up. The sensitivity to irradiation decreases with increasing carbon content in the polymeric samples; in other words in the sequence: PTFE, PVC, PMMA, PC, PE, PS []. Irradiation processes usually proceed very quickly during irradiation in EM, which means that investigations of polymers can often only be performed on badly damaged molecules. These damaged specimens are frequently well suited for investigating the supramolecular structure (the morphology) because the morphology is often unaltered by molecular processes. This generally holds for amorphous polymers, since image formation is based on differences in mass thickness (e.g. between different polymer phases, particles, domains). For semicrystalline polymers, however, the crystalline structure should also be investigated using diffraction contrast, which disappears upon molecular damage. The damage to the specimen can be reduced by taking precautions with the instrumentation and the manner of operation: . If possible, “low-dose” techniques should be used [], which means focussing on one place on the sample and taking micrographs of a different (i.e. previously nonirradiated) place. Low-dose techniques involve performing adjustments to parameters such as the magnification, matching the dosage to the sensitivity of the viewing medium, and focussing on an area of the specimen adjacent to the feature. The beam is then moved with the beam deflection coils, and the shutter is closed to reduce the irradiation of the site adjacent to the feature. The beam is directed onto the feature only when it is being recorded. . The presence of evaporated thin carbon layers on both sides of the specimen improves the conductivity and can reduce specimen movements and the volatilisation of molecular fragments [, ]. The loss of contrast is small, and the increase in dosage has been shown to be as much as tenfold for organic crystals []. . The sensitivity of the photographic material can be enhanced by using special emulsions with a higher silver halide content or special development techniques [, ], or the use of electron image intensifiers with coupled image recording can be helpful. . The application of a higher accelerating voltage for the electrons will lead to a reduction of the inelastic scattering cross-section. The corresponding enhancement of the radiation resistance upon changing from  to  kV is about %, and it increases about threefold upon changing from  kV to  MV [–]. (However, it should also be remembered that the exposure of the photographic film depends on ionisation process, and so the exposure time also increases). . In a HVTEM with an accelerating voltage of  kV or more, the high energies of the electrons allow the use of highly sensitive photographic films, such as special X-ray films with a double coating of thick layers of emulsion and a higher

8.3 Low Contrast of Polymers

177

silver halide content []. Such films cannot be used in a usual TEM because of the lower penetration power of the – kV electrons. . The use of a scanning transmission electron microscope (STEM) instead of a fixed-beam microscope can reduce the damage somewhat []. . By cooling the specimens to low temperatures (cryomicroscopy), the mobility of the polymer molecules and all secondary processes (e.g. mass loss, amorphisation rate of crystals, crosslinking) can be reduced. Cooling specimens to the temperature of liquid nitrogen only has a small effect [, ], but cooling to liquid helium temperatures yields a significant increase in the polymer lifetime [, , ]; see also Sect. ... However, it is still quite difficult to use liquid helium in a special cryomicroscope, and this rules it out as a routine operation for polymer samples. On the other hand, it is also possible to make use of the irradiation sensitivities of polymeric materials for special effects of contrast development: – In the case of polymer blends, the different sensitivities of the polymeric components can result in a stronger mass loss in one component than in another, yielding a the development of contrast at the start of irradiation in the EM. An example is shown in Fig. .. – In the case of semicrystalline polymers, the primary irradiation processes (rupture of chemical bonds, ionisation, etc.) are the same, but the secondary processes of crosslinking may be stronger in the amorphous regions than in the crystalline ones. This gives rise to some other effects, and the result is the development of contrast between the amorphous and crystalline parts; see Sect. .. with Figs. .–.. As mentioned above, the irradiation-induced changes (at the macromolecular level) often do not impede morphological investigations performed at the supramolecular level. Moreover, the morphologies of specimens can be stabilised by applying chemical fixation and staining treatments – essentially by crosslinking the macromolecules and incorporating atoms of heavy elements – which is often necessary when preparing ultrathin sections for TEM or surfaces for SEM investigations. Another common and serious problem is poor performance during in situ deformation tests, although this problem can be overcome by implementing several precautions; see Chap. . The total avoidance of any irradiation-induced effects is only possible if replicas of the materials of interest are investigated (see Sect. .) or if an AFM is used.

8.3 Low Contrast of Polymers A second problem associated with the direct investigation of polymers by transmission electron microscopy (TEM) is the low contrast between structural details. However, there are several methods that can be used to enhance the contrast:

178

8 Problems Associated with the Electron Microscopy of Polymers

. The chemical staining of details can be achieved through the selective incorporation of heavy elements. Different structural details (lamellae, amorphous regions, interfaces, regions of different molecular packing densities or different free volumes, several polymer phases, and others) possess different reactivities to staining media. Also, different staining media are applicable to different materials (e.g. osmium tetroxide, ruthenium tetroxide, phosphoric-tungsten acid, chlorosulfonic acid). . In addition to the commonly used method of chemical staining, there are some physical methods that enhance the contrast between structural details. One method is based on the abovementioned different irradiation sensitivities of different polymeric components. Irradiation-induced changes can also enhance the contrast of semicrystalline polymers; for instance the structure and lamellar arrangements inside spherulites. Another physical method involves developing the contrast between different polymeric components by mechanically straining the sample (this is known as “straining-induced contrast enhancement”). . Structures at surfaces can be “developed” by chemical or physical etching methods. The contrast of structural details can also be enhanced by evaporating metal atoms at small angles (shadowing). These chemical and physical methods of contrast enhancement will be discussed in detail in Chap. , and the methods of structural development at surfaces are explored in Sect. ..

8.4 Methods of Investigating the Morphologies of Polymers 8.4.1 Powders, Particles, Fibres Powders, other small particles or fibres can be investigated “as is”, i.e. without any additional treatment. After mounting the material on a sample holder, its shape, size, and surface structure can be studied directly in SEM or ESEM. Dispersions of very fine particles or fibres can be mounted on a (C-coated) grid for direct TEM inspection. Embedding and ultramicrotomy of the particles is usually needed to reveal the internal particle structure in TEM (see Chap. ). 8.4.2 Bulk Polymers The sizes of the supramolecular structures or morphological features of bulk polymers range from the smallest details that are just above the molecular level (smaller than  nm) up to structures that are larger than  μm – a range of more than five orders of magnitude. Several different preparation techniques and microscopic methods are available for studying structure. However, there is a general methodology for structure determination that should be applied to all microscopic investigations. The various steps included in preparation and investigation methods, as well as the corresponding results obtained, are schematically illustrated in Fig. ..

8.4 Methods of Investigating the Morphologies of Polymers

179

Fig. 8.1. Illustration of the steps involved in sample preparation and investigation methods, and the corresponding results from them during structure determination

Preparation methods for investigating the morphologies of bulk polymers differ from the techniques used for inorganic materials. The usual technique applied to inorganics – chemical or electrolytical thinning – is not applicable to polymers because they swell in solvents and therefore cannot be thinned continuously. Three methods for investigating the morphologies of polymers are generally available (Fig. .): . The preparation of special surfaces (e.g. brittle fracture surfaces, smooth and selectively etched surfaces) that yield information on the internal structure of the material. These surfaces are investigated by means of replicas in the TEM or directly in the SEM or AFM.

180

8 Problems Associated with the Electron Microscopy of Polymers

. The preparation of thin sections by ultramicrotomy, generally after special fixation and staining procedures have been performed. Investigations are carried out by conventional TEM, HVTEM or AFM. . The preparation of special thin films [solution-cast films, focussed ion beam (FIB) sections] with an additional staining treatment (as used for ultramicrotomy) and then studying by TEM or AFM. The methods used to prepare surfaces are covered in the next chapter, while those used to prepare ultrathin or semi-thin sections are discussed in Chaps.  and  and the preparation of special thin films is reviewed in Chap. . These three groups of preparation techniques are generally applicable to various polymers; however, some are more convenient for some groups of polymers while others are better for other polymer groups. A rough summary of the applicability of the preparation techniques to various types of polymers is provided in Table .. Whether or not these different preparation and investigation methods are successful depends on the kind of sample under examination and its structural details, and they do not work to the same extent for all morphological types. Therefore, to perform a full structural determination it is often necessary to apply several techniques. This approach is illustrated for semicrystalline polymers in Fig. .. Additionally, different techniques enable different maximum resolutions in EM. Figure . shows resolution that can be obtained for semicrystalline, lamellar polyethylene when it is prepared with different techniques and contrast enhancements. The structures observed in electron micrographs will usually be estimated qualitatively or semiquantitatively in order to derive information about shapes, distributions, arrangements or types of structural components. Quantitative information about structural details, such as the diameters of particles, the thicknesses of lamellae, or others, are usually determined directly from the micrographs using a PC. The automatic analysis of structural details is covered in the discussion of image processing in Chap. .

8.5 Methods for Studying Micromechanical Processes Due to the variety of different structural details that can occur in polymers, there are also a wide variety of micromechanical processes that can appear under load. These include changes to individual macromolecular segments (on a nanometer scale), localised plastic yielding in the form of crazing or shear bands (at the micrometre scale), up to crack propagation and macroscopic fracture (at the millimetre scale); see Fig. .. Therefore, different techniques for studying these processes are required, which are discussed in detail in Chap. . There are additional advantages of using micromechanical testing for morphology determination: – Components with different mechanical properties can show enhanced contrast due to their different deformabilities (e.g. soft, rubber-like particles in a stiffer

8.5 Methods for Studying Micromechanical Processes 181

Fig. 8.2. Overview of the preparation techniques and electron microscopic methods often used to investigate the morphologies of bulk polymers

182

8 Problems Associated with the Electron Microscopy of Polymers

Table 8.1. Summary of the specimen preparation techniques used for EM investigations of the morphologies of different classes of polymers

x

x

x

x

x x

x x

x x

x

Semicrystalline polymers

Composites

Preparation of “structure developed surfaces” and study of (single-stage or two-stage) replicas by TEM, studying directly by ESEM, AFM or after coating by SEM (see Chap. 9) a) Surfaces “as is” if they show a structure, e.g. after free crystallisation of films or foils from melt or solution b) Smooth (polished or sectioned) surfaces and selective etching (chemically using solvents or physically using ion beams or glow discharge) c) Fractured surfaces from low-temperature fracture, brittle fracture d) Surfaces from “melt fracture” (high-temperature fracture, soft matrix fracture) 2. Preparation of ultra- or semi-thin sections by (cryo)ultramicrotomy after chemical or physical fixation and study by TEM, STEM, HVTEM or AFM (see Chaps. 10, 13) a) Sufficient density differences exist for contrast between structural details b) Enhancement of contrast by chemical treatment (with one chemical agent, by combined attack of several media, or after physical activation) c) Enhancement of contrast via physical effects – irradiation-induced contrast enhancement – straining-induced contrast enhancement d) Use of diffraction contrast, preparation of thin sections with cryoultramicrotomy 3. Preparation of special thin films with additional staining procedure (see Chap. 11) a) Solution-cast films b) Thin sections cut with focussed ion beam (FIB)

Polymer blends, Block copolymers

Applicable for Amorphous polymers

Preparation technique and EM method

1.

x x

x

x

x x

x x

x

x

x x

matrix). This effect is termed “straining-induced contrast enhancement” (see Sect. ..). – The strengths of interfacial layers in heterogeneous polymers can be checked through straining. In particular, a low interfacial strength will result in interfacial rupture (phase decohesion, unbonding, see Fig. .) under strain. This can allow, for instance, the effects of using surface treatments or coupling media to be evaluated.

References

183

References 1. Grubb DT () J Mater Sci : 2. Reimer L () Review of the radiation damage problem of organic specimens. In: Siegel BM, Beaman DR (eds) Electron microscopy. Wiley, New York, p  3. Glaeser RM () Radiation damage and biological electron microscopy. In: Siegel BM, Beanan DR (eds) Electron microscopy. Wiley, New York, p  4. Misell DL (ed) () Developments in electron microscopy and analysis (Conf Ser ). Institute of Physics, Bristol, UK 5. Michler GH () Appl Spectrosc Rev : 6. Fischer EW () Proc th Int Conf Phys Non-Cryst Solids, Clausthal-Zellerfeld, Germany, – Sept , p  7. Baier P et al. () Proc th Eur Congr Electron Microscopy, Vol , The Hague, The Netherlands, – Aug , p  8. Dubochet J, Knapek E, Dietrich I () Ultramicroscopy : 9. Fryer JR, Holland F () Ultramicroscopy : 10. Boudet A, Kubin LP () Proc th Int Congr Electron Microscopy, Vol , Toronto, Canada, – Aug , p  11. Parsons DF, Marko M, King MV () J Microsc : 12. Thomas EL, Humphreys, CJ, Duff WR, Grubb DT () Radiat Effects : 13. Thomas EL, Ast DG () Polymer : 14. Martinez JP et al. () Ultramicroscopy : 15. Michler GH, Dietzsch Ch () Cryst Res Technol : 16. Krause SJ, Allard LF, Bigelow WC () Proc th Int Congr Electron Microscopy, Vol , Toronto, Canada, – Aug , p  17. Glaeser RM, Taylor KA () J Microsc : 18. Dorset DL, Zemlin F () Ultramicroscopy :

9 Preparation of Surfaces

This chapter covers various preparation steps, techniques and results concerning the imaging of surfaces of polymers. Firstly, examples of the direct imaging of surfaces of polymeric samples like powders, fibres, polymer latices, etc. that do not require chemical or physical pretreatment are mentioned. Secondly, chemical etching procedures like permanganic etching and physical etching techniques like plasma or ion etching are described, and the results of these procedures are discussed for several groups of polymers like polyolefins, block copolymers, and others. Thirdly, the preparation of fracture surfaces is explained and illustrated, including cryofracture at low temperatures and the special method of soft matrix fracture or melt fracture at elevated temperatures. Finally, examples of the preparation of surface replicas that can be imaged in the TEM are presented.

9.1 Overview Figure . in the preceding chapter showed that the preparation and investigation of surfaces is one of the ways in which the morphologies of polymers can be studied. The external surfaces of polymeric solids only contain any information on the internal morphology in some cases, because they are usually strongly modified by the processing conditions, e.g. pressure or injection moulding, extrusion, or others. Therefore, special “structured surfaces” from the interior must be prepared, using different etching techniques or fracture processes. These different techniques are discussed in the following sections.

9.2 Surfaces in Their Natural Form Surfaces only reveal a polymeric morphology in some cases, such as for foils or fibres that were freely crystallised from the melt or solution. Another example is powder obtained directly from the crystallisation process, where the surface morphology is of great research interest (see the UHMWPE powders in Fig. .). In the simplest cases, these “as is” surfaces can be mounted onto a sample holder and studied directly by ESEM or, after evaporating a thin conductive layer, by typical SEM. To avoid agglomeration in the case of fine powders, the particles should be well separated in a liquid suspension or through ultrasound treatment. Particles that

186

9 Preparation of Surfaces

are small enough, i.e. that have diameters that are smaller than about  nm, can be dispersed onto a carbon-coated grid for TEM inspection. Latex particles, for instance, can be advantageously sprayed onto a grid from the dispersion in an ultrasound bath. Latex particles of hard polymers, e.g. PS particles, are stable enough to maintain their spherical shape, whereas rubber latex must be stabilised by chemical fixation before spraying (e.g. polybutadiene rubber particles can be fixed with OsO or bromine). By the way, latex particles of PS with exactly the same molecular weight and therefore the same diameter can be prepared, and these can be used as a simple way to check the magnification of a TEM.

9.3 Smooth and Etched Surfaces Smooth internal surfaces of bulk polymers can be prepared by polishing [] or sectioning in an ultramicrotome (Chap. ). Very smooth surfaces prepared by applying ultra- or cryoultramicrotomy of semicrystalline to heterogeneous polymers are suitable for AFM inspection. If the surfaces are not flat enough, they can be pressed against a really smooth surface, e.g. a mica cleavage surface at higher temperatures. Moreover, a parallel alignment of the sample surface in relation to the scanner is desirable for successful AFM imaging. Deviations from a very flat and well-aligned surface can lead to erroneous interpretations of AFM data (see Chap. ). Smooth surfaces of particle-filled polymers (composites) can be studied in SEM with backscattered electrons, i.e. in material contrast, revealing the size, shape and distribution of the filler particles. However, the preparation of brittle fracture surfaces (see below) gives better results. For polymer blends, one polymer phase can be chemically stained (e.g. rubber particles in high-impact polymers can be stained with osmium tetroxide), thus giving a sufficient material contrast in SEM (using backscattered electrons); one example is shown in Fig. .. In all other cases, the internal polymer structure must be “developed” on the surface via selective etching processes, which can be performed chemically with solvents, acids, and bases and physically using ion etching or glow discharge. Etching, in general, is a process that erodes material from the surface of a sample, changing surface properties like the topography, wettability, functionality, optical qualities, etc. In this context, etching is employed as an industrial process for material modification. On the other hand, chemical etching procedures have long been used to facilitate metallographic texture analyses. Etching for contrast enhancement in polymer microscopy includes the removal of one or more components (or “phases”) of the polymer, polymer blend or composite by physical or chemical means. In most cases, the constituents of a heterogeneous polymer system are more or less sensitive to a certain treatment. The idea is to produce a surface topography that can be utilised for image contrast formation using secondary electrons in the SEM or ESEM. Etching procedures can be applied to original sample surfaces as well as to fracture surfaces or block faces smoothed by a microtome.

9.3 Smooth and Etched Surfaces

187

9.3.1 Chemical Etching The most common etching method is the selective elimination of a material component through the application of selective solvents, strongly oxidising acids, basic compounds, or mixtures of these substances [–]. Excellent results are achieved, for instance, by applying a mixture of water, sulfuric acid, orthophosphoric acid and potassium permanganate to semicrystalline polyolefins [,]. Essentially, the etching effect is due to the higher sensitivity of the amorphous, less ordered portion of the material to the strong oxidising agent. Differences in the degradation rate produce a relief, creating “mountains” on the crystalline material; spherulites and stacks of crystalline lamellae become visible, as do fibrillar structures of oriented fibres [–]. In rubber-toughened polymers, rubber particles can be removed by the application of a proper etching agent. For instance, a mixture of sulfuric acid, orthophosphoric acid, water and chromic acid is used to remove the rubber phase []. For optimal results, it is necessary to find the appropriate etching agent, and one must balance (at least) the following parameters: composition and concentration of the etchant, temperature, and duration. The sample should be agitated vigorously using a magnetic stirrer or a shaker to remove debris from the surface and to avoid the redeposition of low molecular weight degradation products. Some authors use ultrasonic cleaners to achieve better etching results. In any case, subsequent rinsing and drying steps will be necessary. The following example will elucidate these processes. This etching procedure, commonly denoted “permanganic etching”, follows the recipes provided by Olley et al. that have been applied with great success to a number of polyolefins and polyolefin blends [–,]. The somewhat simplified procedure described here is easy, fast and safe [,]. It can easily be modified to meet changing requirements when material parameters (e.g. crystallinity, crystalline modifications) are varied. For plate-pressed or injection-moulded standard iPP in the crystalline α-form, the procedure is as follows: – Cut a sample of appropriate size from the region of interest. – Produce a flat surface using a microtome. If possible, a cryomicrotome should be used to avoid plastic deformation of the polymer. – Prepare the etchant:  ml of distilled water are placed into a flask that contains a Teflon-coated magnetic stirrer. Add  ml of concentrated sulfuric acid slowly. Add . g of potassium permanganate powder slowly under continuous stirring. This work must be done under a ventilated hood. Wear gloves and eye protection! Do not try to produce more of the etchant; it will degrade quite quickly when stored in the laboratory. There is a risk of explosion if the mixture is created in the wrong way. The proper etchant has a deep olive green colour. If it is used it becomes violet or brown. – Place the sample in a test tube, add the etchant using a pipette, and shake it well for  min. – Remove the sample and rinse it with distilled water. Put the sample in another (clean) test tube, add water, and place it in an ultrasonic bath for several minutes. – Place the sample on filter paper and allow it to dry.

188

9 Preparation of Surfaces

Fig. 9.1. Spherulitic morphology of α-iPP after permanganic etching. SEM image

Fig. 9.2. Morphology of a blend of PP and PEO after permanganic etching. PEO phase is selectively removed, and the spherulitic morphology of PP is visible. SEM image

9.3 Smooth and Etched Surfaces

189

Of course, the rinsing procedure can also be modified. For instance, there can be an additional rinsing step using acetone. Alternatively, the procedure also can be applied to oriented fibres, cryofracture surfaces or free surfaces produced by solution-casting techniques. Some examples of surfaces etched in the described manner are given in Figs. . to .. In Figs. . and ., the spherulitic morphology of α-iPP is transformed into a surface topography that is detectable using the secondary electron signal of the SEM. Moreover, the amorphous PEO phase of a PP/PEO blend (Fig. .) is removed. More images of etched PP samples are reproduced in Chap. . In Fig. . the same procedure is applied to a SBS block copolymer. The result is a nanoscopic surface topography arising from the typical phase nanostructure of the SBS. The same etchant is used for syndiotactic PS (Chap. ), ethylen-octen copolymers (Chap. , Fig. .), and rubber-toughened PP (Chap. , Fig. .), where only the etching times have been changed. Table . provides an overview of selected polymers and etching agents that have been used successfully. Besides literature references, references for the examples that are discussed in this book are provided.

Fig. 9.3. Nanostructured surface of SBS block copolymer after permanganic etching. The polybutadiene phase is selectively removed. ESEM image

190

9 Preparation of Surfaces

Table 9.1. Etching techniques for selected polymers Polymer

Etching agent, action, result

Example in Chapter

References

– permanganic – solvent (xylol) amorphous phase removed; spherulitic texture, lamellar structure – permanganic amorphous phase removed; spherulitic texture, lamellar structure – permanganic amorphous matrix degraded, crystalline lamellae visible – permanganic amorphous phase removed; spherulitic texture – permanganic removal of the amorphous or less crystalline phases

16

[5–7, 12–16]

– methylamine – alcohols amorphous phase removed; spherulitic texture, lamellar structure – acids amorphous phase removed; spherulitic texture – hydrolysis – nitric acid fibrillar structure of oriented fibres – solvents: heptane, benzol, xylol; amorphous phase removed; spherulitic texture, lamellar structure – diethylamine, ethanol amorphous phase removed; spherulitic texture, lamellar structure (if applicable) – alcohols amorphous phase removed; spherulitic texture

23

Polyolefins PE

PP

Copolymers of polyolefins

EOC

COC

Blends of polyolefins

16

5

[17]

16

17, 18

Polyesters PHA

PLA

PET

PA

PC

POM

[18, 19]

23

9

[20, 21]

[22, 23]

[24]

[19]

9.3 Smooth and Etched Surfaces

191

Table 9.1. (continued) Polymer

Etching agent, action, result

Example in Chapter

– permanganic PEO phase removed amorphous phase of PP removed; spherulitic texture, lamellar structure – permanganic PET phase removed amorphous phase of PP removed; spherulitic texture, lamellar structure – chromic acid – permanganic rubber phase removed – chromic acid – permanganic rubber phase removed

9

References

Blends PP/PEO

PET/PP

HIPS

ABS

PCL/PS PE/NR

9

[8]

[8]

[25] [26]

– permanganic

Block copolymers SBS

– permanganic butadiene phase removed

9

Others PVDF

PEEK

sPS

– phosphorus pentoxide, – chromium trioxide – permanganic amorphous phase removed; spherulitic texture, lamellar structure – permanganic amorphous phase removed; spherulitic texture, lamellar structure – permanganic amorphous phase removed; spherulitic texture, lamellar structure

[27, 28]

9

[29]

16

9.3.2 Physical Etching Another way to achieve topographic contrast on polymer surfaces is physical etching using gaseous ions. This treatment usually takes place in a vacuum chamber where argon ions are generated by cascade ionisation in a low-pressure vacuum (.– mbar). This ion plasma procedure is similar to the process used in the sputter coater to create specimen surfaces conducive to SEM (Chap. ). In contrast to a sputter coater in a plasma etching device, here the specimen itself acts as the target. The plasma etching can be performed either in a directly coupled electric field for anisotropic etching, e.g. at a certain angle to generate a well-defined structure on

192

9 Preparation of Surfaces

Fig. 9.4. SEM micrographs of high-frequency plasma-etched surfaces of UHMWPE powder particles. Left: etching by Ar ions; right: etching by O ions. Conditions: pressure 2.3  10−1 mbar, frequency 40 kHz , plasma power 40 W, etching time for argon ions 30 min and for oxygen ions 40 min

a specimen surface, or in a high-frequency electric field for isotropically etching the surface structures of the specimen. The latter is a very sparing and sensitive etching technique in comparison to the chemical one mentioned above. On the other hand, a high-frequency plasma treatment can also be carried out by reactive ions like oxygen, chlorine or fluorine; among these oxygen is the most important gas in plasma processes. The application of oxygen [–] leads to oxidative reactions with the organic molecules of the polymer that generate O radicals that are able to crack hydrocarbon chains and to oxidise them to CO and H O. Due to the fact that different polymeric structural details (crystalline and amorphous parts, different polymer phases, inorganic particles) have different etching rates, the result of plasma etching can be the elucidation of the structural details, depending on the etching time. However, the etched surfaces only represent nearsurface structures during the initial stages of etching. At advanced stages, i.e. after longer etching times, special etch figures that are more influenced by the etching conditions often occur []. In Fig. . two examples of high-frequency plasma etching by argon and oxygen ions of a reactor powder of ultrahigh molecular weight polyethylene (UHMWPE) are demonstrated.

9.4 Fracture Surfaces The simplest way to study multiphase, heterogeneous materials is to produce brittle fracture surfaces. This is usually done at low temperatures (e.g. at the temperature of liquid nitrogen) to avoid plastic deformation, which would hide the morphology. The fracture path occasionally follows structural details, e.g. spherulites in semicrystalline polymers, phase boundaries, or interfaces in polymer blends, so that they become visible in SEM. The application of the etching techniques discussed in Sects. .. and .. occasionally enhances structure visibility. Fracture surfaces of polymer composites often clearly reveal the size, shape and distribution of the inorganic filler particles in a polymer matrix. Figure . shows a particulate-filled PMMA

9.4 Fracture Surfaces

193

Fig. 9.5a,b. SEM images of the same area of a carbon-coated fracture surface of a particulate-filled PMMA, used as bone cement (see Sect. 23.3.3): a SE image highlighting the topography of the fracture surface; b BSE material contrast image showing the distribution of the inorganic filler in the PMMA matrix

used as bone cement (see Sect. ..) in SEM micrographs with secondary electrons (a) and with backscattered electrons, i.e. in material contrast (b). Using EDXA, the chemical nature of the filler particles is made detectable. Brittle fracture paths do not always follow distinctive structures, particularly if there are only small differences in the mechanical stiffnesses of the individual components in the frozen state or if there are no clearly marked interfaces between components. Therefore, different accidental features are often created on the surfaces, which vary with the fracture process itself, e.g. crack velocity. These accidental features make the identification of real structures somewhat difficult. This problem can be partly overcome by using an unusual technique to produce a fracture surface at higher temperatures, e.g., in the melting range of the matrix polymer (called “melt fracture” or “soft matrix fracture” [, ]). Using this technique, hard filler particles in a matrix become more clearly detectable, as illustrated in Fig. .. Another example of this unconventional way to prepare fracture surfaces is a carbon black modified rubber fractured at   C, see Fig. .. This unconventional “soft matrix fracture” is based on a specific property of polymer materials, namely the (up to three-decade) difference in the Young’s modulus below and above the glass transition temperature. In this approach, the fracture surface must be obtained at a temperature at which the matrix is soft (the temperature must be sufficiently high above the glass transition temperature of the matrix) and the inclusions are hard (below the glass transition temperature of the particles) at the same time []. The temperature of fracture must be in a range ΔT where the matrix has a low modulus and the inclusions have a high modulus; see Fig. .. This difference in moduli can be obtained by varying the sample temperature. Besides the study of polymer composites, this procedure can also yield successful results for polymer blends, provided that a sample state can be reached in which the matrix is rubber-like while the particles are hard. Such an example was presented previously in [] with an ABS polymer, where the soft rubber particles were hardened by osmium tetroxide treatment and the sample was then notched and fractured at   C. This temperature is well above the glass transition temperature of the styrene–

194

9 Preparation of Surfaces

Fig. 9.6a,b. Visibility of inorganic filler particles (CaSO4 and TiO2 ) in a PP matrix on fracture surfaces in SEM, using different fracture processes: a usual brittle low-temperature fracture; b high-temperature fracture, prepared at 140  C (“soft matrix fracture”). From [35], reproduced with the permission of Chapman and Hall

Fig. 9.7. A sketch of the moduli (G) conditions needed to apply the method using soft matrix fracture in the temperature range ΔT (after [35])

acrylonitrile matrix, but the stained particles remain hard, so that the conditions for the moduli (Fig. .) are fulfilled. Please note that the same staining of rubber particles is used to visualise the particles on cut surfaces in SEM using material contrast

9.5 Investigation of Surfaces

195

with backscattered electrons (see Fig. .) and in thin sections in TEM (see Figs. . and .). In general, methods for the preparation and electron microscopic investigation of surfaces (etched surfaces or fracture surfaces) are relatively easily and successfully performed if the polymers contain large structures or structurally clearly distinguishable parts. They are unsuited to revealing very fine details or for examining the surface at high resolution. There are two other surface techniques, which are only interesting from a historical point of view: – The method of decorating the surface with gold, i.e. the evaporation of Au under high vacuum onto substrates, is very useful for studying surface steps of monoatomic height on inorganic crystal surfaces []. However, there is no clear decoration effect on polymer surfaces. Another problem arises from the need to cover the gold particles with a carbon film and to remove film and particles from the substrate, which is very difficult for polymers (see below). There are reports of the decoration of ultrathin solution-cast films of PE and PA [, ] and of the decoration of polymer crystals with paraffin []. – Another rarely used technique with a highly variable success rate is called “surface rupture”. When a matrix material is peeled off a polymer’s surface, especially after activation by oxygen etching, a thin polymer layer can be torn from the polymer surface and directly studied by TEM [].

9.5 Investigation of Surfaces Polymer surfaces exhibiting structural development after etching or fracture are usually studied directly by SEM or ESEM. SEM investigation demands that the surface is covered with a very thin conductive layer, which is usually performed by evaporating carbon or, for a better contrast, carbon/gold under a high vacuum or by a sputtering process. In the past, surfaces have been investigated by TEM using the replica technique. A “single-step replica” is produced by evaporating a thin carbon film onto the surface followed by platinum/carbon shadowing (see below) and partial dissolution of the matrix. This film with platinum structures on original surface edges gives an impression of the surface and can be investigated in the TEM. Since polymers usually swell strongly during dissolution, a thin evaporated carbon film will usually be destroyed. Therefore, a “two-step replica” can be used instead. In a first step, a primary replica is made of the surface by means of a thick, mechanically stable plastic foil (matrix). Next, after the foil has been mechanically removed from the surface, a secondary replica is made of the contact plane of the latter, e.g. by using a platinum/carbon shadow cast (see below). After dissolving the matrix foil, the C-film is investigated in the TEM. Matrix materials that are particularly well suited to this are the solvoplastic substances acetyl cellulose (“Bioden”, soluble in methyl acetate) and cellulose acetobutyrate (“Triafol”, soluble in acetone), which, after slight dissolution of their

196

9 Preparation of Surfaces

contact surfaces, are pressed as a thin foil onto the specimen to be replicated. Polymer solutions such as cellulose acetate in acetone and polymerising substances, or, as a special technique, an evaporated silver matrix reinforced by an electrolytically produced stable copper layer, are also suitable. Polymer residues that adhere to the matrix foil can be removed by activated oxygen plasma or by thermal decomposition in a high vacuum []. To enhance the contrast of the structures, shadowing is carried out in a vacuum evaporator using a heavy metal, such as gold, platinum or a platinum/palladium mixture. Of the numerous replica techniques developed for electron microscopy, the simultaneous evaporation of platinum/carbon film – nm thick [] has proven to be the best. The evaporation of the metal is performed at an oblique angle to the surface (generally at an angle of   C to   C). The edges of the structure facing away from the source of the heavy metal will not be coated. The length of this uncoated area (or shadow) of carbon film depends on the evaporation angle and the height of the structure.

References 1. Linke U, Kopp WU () Prakt Metallogr : 2. Berndtsen N, Jansen FJ () Präparation von Polymeren für die Licht- und Elektronenmikroskopie. Institut für Kunststoffverarbeitung an der RWTH, Aachen, p  3. Olley RH () Sci Progr : 4. Sawyer LC, Grubb DT () Polymer microscopy, nd edn. Chapman and Hall, London, p  5. Olley RH, Hodge AM, Bassett DC () J Polym Sci Polym Phys : 6. Olley RH, Bassett DC () Polymer : 7. Bassett DC, Olley RH () Polymer : 8. Bucknall CB, Drinkwater IC, Keast WE () Polymer : 9. Olley RH, Bassett DC, Hine PJ, Ward IM () J Mater Sci : 10. Henning S, Michler GH, Ania F, Baltá Calleja FJ () J Colloid Polym Sci : 11. Henning S, Michler GH () In: Michler GH, Baltá Calleja FJ (eds) Mechanical properties of polymers based on nanostructure and morphology. CRC Press, Boca Raton, FL, p  12. Mercx FPM, Benzina A, van Langeveld AD, Lemstra PJ () J Mater Sci : 13. Li J, Lee YW () J Mater Sci : 14. Shahin MM, Olley RH, Blissett MJ () J Polym Sci Polym Phys : 15. Zok F, Shinozaki DM () J Mater Sci Lett : 16. Reding FP, Walter ER () J Polym Sci : 17. Doshev P, Lohse G, Henning S, Krumova M, Heuvelsland A, Michler G, Radusch HJ () J Appl Polym Sci (): 18. Organ SJ, Barham PJ () J Mater Sci Letters : 19. Shahin MM, Olley RH () J Polym Sci Polym Phys : 20. Cagiao ME, Baltá Calleja FJ, Vanderdonckt C, Zachmann HG () Polymer (): 21. Yoshioka T, Okayama N, Okuwaki A () Ind Eng Chem Res : 22. Kallweit () Progr Coll Polym Sci : 23. Bartosiewics L () J Polym Sci Polym Phys : 24. Kämpf G () Kolloid-Zeitschrift (): 25. Shabana HM, Olley RH, Bassett DC, Jungnickel BJ () Polymer : 26. Hudec I, Sain MM, Kozankova J () Polym Test : 27. Vaughan AS () J Mater Sci : 28. Hsu TC, Geil PH () J Mater Sci : 29. Olley RH, Basset DC, Blundell DJ () Polymer : 30. Spit BJ () Polymer : 31. Spit BJ () Faserforsch Textiltech :

References

197

32. Kämpf G () Progr Colloid Polym Sci : 33. Friedrich J, Gähde J () Acta Polym : 34. Lednicky F () In: Sedlaèek B (ed) Morphology of polymers. Walter de Gruyter & Co, Berlin, p  35. Lednicky F, Michler GH () J Mater Sci : 36. Lednicky F, Pelzbauer Z () Polym Test : 37. Bethge H, Heydenreich J (eds) () Electron microscopy in solid state physics. Elsevier, Amsterdam 38. Bassett GA, Blundell DJ, Keller A () J Macromolec Sci Physics B: 39. Spit BJ () J Macromol Sci Phys B : 40. Wittmann JC, Lotz B () J Polym Sci Polym Phys : 41. Preuß HHW () Plaste und Kautschuk : 42. Marichin VA, Mjasnikova LP () Prib Tekh Eksp No : 43. Bradley DE () Brit J Appl Phys :

10 Preparation of Thin Sections: (Cryo)ultramicrotomy and (Cryo)microtomy

Microtomy and ultramicrotomy are standard methods used for the preparation of ultrathin or semi-thin sections and flat surfaces of plastics, biological and biomedical objects. Brief discussions of the instrumentation required for these techniques, of working with a microtome and ultramicrotome, of specimen preparation (embedding, trimming, fixation), of sectioning (wet and dry sectioning, sectioning at room or cryo temperatures) and of modern trends in this specific field are presented in this chapter. In particular, the handling of the devices in practice and the influence of the sectioning parameters are described and illustrated using many examples from polymer research. In the final section, typical errors encountered during the different working steps and solutions to them are discussed in detail.

10.1 Overview Ultramicrotomy (including cryoultramicrotomy) is a standard method for the preparation of ultrathin/semi-thin sections as well as very flat surfaces of plastics, biological and biomedical objects for various microscopic investigations. Improvements in preparation techniques over the last few decades have demonstrated that thin sections of different materials that are free from artefacts can be successfully prepared for electron microscopic investigations. Therefore, successful sectioning now depends primarily on the experience of the experimentalist rather than on the instrumentation used. In order to avoid sectioning-induced errors and to fully exploit the capability of an ultramicrotome, one must master the optimum specimen preparation and sectioning technique. The traditional application of ultra- and cryoultramicrotomy involves the sectioning of soft materials such as the abovementioned biological or biomedical substances and polymers. Recent experiences demonstrate, however, that even hard materials can be successfully ultramicrotomed. In addition to soft metals like aluminium (Al) or copper (Cu), materials as hard as steel, ceramics and even very hard substances like sapphires and carbides have been successfully sectioned by ultramicrotomy. These developments show that the technique of ultramicrotomy can be expected to extend to other sectors of materials science, supplementing existing preparation techniques [–].

200

10 Preparation of Thin Sections: (Cryo)ultramicrotomy and (Cryo)microtomy

10.2 Instrumentation Depending on the section thickness and sectioning temperature, the following instruments are necessary. 10.2.1 Microtomes One generally prepares thin ( μm) and semi-thin (.– μm) sections at room temperature with the aid of a microtome. The sections and the flat surfaces of the blocks where the sections were cut can be inspected by means of SEM, AFM and HVTEM. 10.2.2 Ultramicrotomes Using ultramicrotomes, semi-thin (thickness .– μm) as well as ultrathin (thickness